{"object":"list","total":653,"limit":100,"offset":0,"data":[{"id":"allenai/molmoact","name":"MolmoAct","title":"MolmoAct: Action Reasoning Models that can Reason in Space","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"allenai/MolmoAct-7B-D-0812","weights_gated":false,"license_class":"permissive","parameter_count_b":8.119,"vram_min_gb":19.5,"supported_quantizations":["fp32","bf16"],"vram_at_4bit_gb":8.1,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":"transformers","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action","trajectory","latent"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.77,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"published","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open weights, data and training code released by AI2.","vendor":"Allen Institute for AI","weights_url":"https://huggingface.co/allenai/MolmoAct-7B-D-0812","routable":true,"featured":true,"contracts":["act","ground"],"modalities":{"inputs":["rgb","language"],"outputs":["action","trajectory"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2508.07917","code_url":"https://github.com/allenai/MolmoAct","project_url":"https://allenai.org/blog/molmoact","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008978,"per_frame_max":0.015836,"per_action_step_min":0.003142,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"allenai/molmoact\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"allenai/molmoact2","name":"MolmoAct2","title":"MolmoAct2 (base checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"allenai/MolmoAct2","weights_gated":null,"license_class":"permissive","parameter_count_b":5.0,"vram_min_gb":12.0,"supported_quantizations":["fp32","bf16"],"vram_at_4bit_gb":5.0,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":180.0,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":180.0,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":989.4,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.77,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"inferred","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Ai2's MolmoAct2 (May 2026): an open vision-language-action model, a Molmo2-ER backbone joined to a flow-matching action expert by per-layer KV conditioning, about 5B parameters. This is the post-trained, multi-embodiment base checkpoint, which Ai2 describes as \"a foundation checkpoint for further robot fine-tuning\": it is not a ready-to-run policy, and Ai2 publishes no inference code for it. Served on Sidekick for /v1/act on the DROID (Franka) setup, by name only: Ai2's DROID server run against this base checkpoint with the franka_droid norm tag from its own norm_stats.json (deploy/modal/molmoact2_policy.py, 1x L40S), so an exterior and a wrist camera, an 8-wide state and a task in, absolute joint targets with the gripper out. As a base it is not tuned for any one robot; a fine-tune made from it, like Ai2's own DROID fine-tune, is what does best on a robot. Ai2's own ready-to-run fine-tunes are LIBERO, DROID, BimanualYAM and SO100_101. Apache-2.0 per Ai2's repository; Ai2 asks for use under its Responsible Use Guidelines.","vendor":"Allen Institute for AI","weights_url":"https://huggingface.co/allenai/MolmoAct2","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","language","proprio"],"outputs":["action"]},"embodiments":["arm","bimanual"],"action_spaces":["joint_pos_7_grip","joint_pos_14"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2605.02881","code_url":"https://github.com/allenai/molmoact2","project_url":"https://allenai.org/blog/molmoact2","providers":["baseten","fal","modal","modal-molmoact2-base","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.027386,"per_action_step_min":0.002,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"allenai/molmoact2\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_7_grip\",\n)"}},{"id":"allenai/molmoact2-think","name":"MolmoAct2-Think","title":"MolmoAct2 (Think: depth-reasoning base checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"allenai/MolmoAct2-Think","weights_gated":null,"license_class":"permissive","parameter_count_b":5.0,"vram_min_gb":12.0,"supported_quantizations":["fp32","bf16"],"vram_at_4bit_gb":5.0,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":790.0,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":790.0,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":989.4,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.77,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"inferred","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Ai2's MolmoAct2-Think (May 2026, about 5B): MolmoAct2 \"with depth-token reasoning\". Before acting it can predict \"a compact 10 x 10 discrete depth representation\" and condition the action expert on that depth-aware cache. Like MolmoAct2 it is a post-trained, multi-embodiment base checkpoint, \"a foundation checkpoint for further robot fine-tuning rather than ... a ready-to-run policy\", with no inference code from Ai2. Served on Sidekick for /v1/act on the DROID (Franka) setup, by name only: Ai2's DROID server run against this base checkpoint with the franka_droid norm tag from its own norm_stats.json (deploy/modal/molmoact2_policy.py, 1x L40S), so an exterior and a wrist camera, an 8-wide state and a task in, absolute joint targets with the gripper out. As a base it is not tuned for any one robot; a fine-tune made from it, like Ai2's own DROID fine-tune, is what does best on a robot. Ai2's ready-to-run depth-reasoning example is its LIBERO fine-tune (MolmoAct2-Think-LIBERO). Apache-2.0 per Ai2's repository.","vendor":"Allen Institute for AI","weights_url":"https://huggingface.co/allenai/MolmoAct2-Think","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","language","proprio"],"outputs":["action"]},"embodiments":["arm","bimanual"],"action_spaces":["joint_pos_7_grip","joint_pos_14"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2605.02881","code_url":"https://github.com/allenai/molmoact2","project_url":"https://allenai.org/blog/molmoact2","providers":["baseten","fal","modal","modal-molmoact2-think","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.024905,"per_action_step_min":0.002,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"allenai/molmoact2-think\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_7_grip\",\n)"}},{"id":"amap-ml/dreamx-world-1-0","name":"DreamX-World 1.0","title":"DreamX-World 1.0: A General-Purpose Interactive World Model","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"GD-ML/DreamX-World-5B","weights_gated":false,"license_class":"permissive","parameter_count_b":5.28,"vram_min_gb":40.0,"supported_quantizations":["fp32"],"vram_at_4bit_gb":5.3,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":26000.0,"latency_hardware":"H20","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","action"],"modalities_out":["video"],"model_type":"interactive-game-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":1.0,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (GD-ML/DreamX-World-5B).","vendor":null,"weights_url":"https://huggingface.co/GD-ML/DreamX-World-5B","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.16993","code_url":"https://github.com/AMAP-ML/DreamX-World","project_url":null,"providers":["baseten","fal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.011639,"per_frame_max":0.021061,"per_action_step_min":0.004069,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"amap-ml/dreamx-world-1-0\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"baai/emu-3","name":"Emu 3","title":"Emu3: Next-Token Prediction is All You Need","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"BAAI/Emu3-Gen","weights_gated":false,"license_class":"permissive","parameter_count_b":8.49,"vram_min_gb":20.4,"supported_quantizations":["fp32","bf16"],"vram_at_4bit_gb":8.5,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["language","rgb","video"],"modalities_out":["text","video"],"model_type":"video-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (BAAI/Emu3-Gen).","vendor":null,"weights_url":"https://huggingface.co/BAAI/Emu3-Gen","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["latent"]},"embodiments":["agnostic"],"action_spaces":[],"year":2024,"arxiv_url":"https://arxiv.org/abs/2409.18869","code_url":"https://github.com/baaivision/Emu3","project_url":"https://emu.baai.ac.cn","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008811,"per_frame_max":0.016716,"per_action_step_min":0.003084,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"baai/emu-3\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"berkeley/ego-vcp","name":"Ego-VCP","title":"Ego-Vision World Model for Humanoid Contact Planning","domain":"embodied","domains":["embodied"],"family":"locomotion","families":["locomotion"],"subsection":"4. World Models for Locomotion","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://github.com/HybridRobotics/Ego-VCP/tree/main/wm_logs","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":25,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["depth","proprio"],"modalities_out":["action","value"],"model_type":"latent-world-model","task_suitability":"locomotion","environment_type":"general","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.62,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (https://github.com/HybridRobotics/Ego-VCP/tree/main/wm_logs).","vendor":null,"weights_url":"https://github.com/HybridRobotics/Ego-VCP/tree/main/wm_logs","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","tactile","proprio"],"outputs":["video"]},"embodiments":["humanoid"],"action_spaces":["joint_pos_23","joint_pos_29","ee_delta_6d_grip"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2510.11682","code_url":"https://github.com/HybridRobotics/Ego-VCP","project_url":"https://ego-vcp.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008439,"per_frame_max":0.015765,"per_action_step_min":0.002954,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":38,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"berkeley/ego-vcp\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_23\",\n)"}},{"id":"black-forest-labs/flux-3-action","name":"FLUX 3 Action","title":"FLUX 3 Action (DROID)","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"black-forest-labs/flux-3-action-droid","weights_gated":true,"license_class":"conditional","parameter_count_b":7.0,"vram_min_gb":32.0,"supported_quantizations":["bf16","fp8"],"vram_at_4bit_gb":7.0,"weights_precision":"bf16","quantized_builds":[{"precision":"fp8","rung":"fp8","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://github.com/black-forest-labs/flux-action/blob/main/docs/setup.md"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":410.0,"latency_hardware":"H200","latency_gpu_count":1,"action_frequency_hz":15,"replan_hz":0.47,"action_chunk_len":32,"action_horizon":32,"time_to_first_action_ms":410.0,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language","proprio"],"modalities_out":["action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"derived","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"derived","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"inferred","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Black Forest Labs' \"7B open weights world action model\" (September 2026), served for /v1/act from its DROID checkpoint black-forest-labs/flux-3-action-droid through BFL's own Python API (deploy/modal/flux3_action_policy.py, 1x H100, about 32 GB in BF16). Three 360x640 cameras, an 8-wide Franka state and a task in; 32 absolute joint targets and the gripper at 15 Hz out, all 32 executed before replanning (BFL's inference docs). The model predicts video internally, but the released inference API returns actions only. Code Apache-2.0; weights under the FLUX Kommunity License v1.0, which allows commercial use of outputs only for users under US$5M annual revenue and otherwise needs a BFL licence. Live on Sidekick: served on Modal (deploy/modal/flux3_action_policy.py, 1x H100), it passed the live probe on 21 Oct 2026 with 32 steps of 8 joints, about 1.0 s warm and 4 min from cold.","vendor":"Black Forest Labs","weights_url":"https://huggingface.co/black-forest-labs/flux-3-action-droid","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","text","proprio"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["joint_pos_7_grip"],"year":2026,"arxiv_url":null,"code_url":"https://github.com/black-forest-labs/flux-action","project_url":"https://bfl.ai/models/flux-3-action","providers":["baseten","fal","modal","modal-flux3-action","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.034895,"per_action_step_min":0.002,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"black-forest-labs/flux-3-action\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_7_grip\",\n)"}},{"id":"cladernyjorn/up-vla","name":"UP-VLA","title":"UP-VLA: A Unified Understanding and Prediction Model for Embodied Agent","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"CladernyJorn/UP-VLA-Calvin","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["action","video","text"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.54,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2501.18867","code_url":"https://github.com/CladernyJorn/UP-VLA","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008767,"per_frame_max":0.017732,"per_action_step_min":0.003068,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"cladernyjorn/up-vla\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/aether","name":"Aether","title":"Aether: Geometric-Aware Unified World Modeling","domain":"general","domains":["general"],"family":"vision-3d","families":["vision-3d"],"subsection":"3. Building World Models from 3D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"AetherWorldModel/AetherV1","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action"],"modalities_out":["video","3d","trajectory"],"model_type":"3d-4d-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"MIT, verified from the GitHub LICENSE file (InternRobotics/Aether) and the HF card. Weights are public and unrestricted: a 5B CogVideoX-based transformer doing action-conditioned prediction, 4D reconstruction and goal-conditioned planning. Served by deploy/modal/aether_world.py.","vendor":"Shanghai AI Lab","weights_url":"https://huggingface.co/AetherWorldModel/AetherV1","routable":true,"featured":true,"contracts":["predict","evaluate"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2503.18945","code_url":"https://github.com/OpenRobotLab/Aether","project_url":"https://aether-world.github.io/","providers":["baseten","fal","modal","modal-aether","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.017772,"per_action_step_min":0.002,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":38,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/aether\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/cellflux","name":"CellFlux","title":"CellFlux: Simulating Cellular Morphology Changes via Flow Matching","domain":"science","domains":["science"],"family":"science","families":["science"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"suyc21/CellFlux","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["tensor"],"modalities_out":["image"],"model_type":"scientific-world-model","task_suitability":"scientific","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.55,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"verified","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://github.com/yuhui-zh15/CellFlux)","vendor":null,"weights_url":"https://huggingface.co/suyc21/CellFlux","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["tensor"],"outputs":["field"]},"embodiments":["none"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2502.09775","code_url":"https://github.com/yuhui-zh15/CellFlux","project_url":"https://yuhui-zh15.github.io/CellFlux/","providers":["baseten","fal","modal","sidekick-cloud","vendor-direct"],"pricing":{"per_frame_min":0.017339,"per_frame_max":0.024061,"per_action_step_min":0.006069,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":146,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/cellflux\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/cosmos","name":"Cosmos","title":"Cosmos World Foundation Model Platform for Physical AI","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"classic","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos-Predict1-7B-Text2World","weights_gated":true,"license_class":"conditional","parameter_count_b":7.0,"vram_min_gb":24.4,"supported_quantizations":["bf16"],"vram_at_4bit_gb":7.0,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-workstation","hardware_rank":3,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 desktop (32GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["language","rgb","video"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":null,"hardware_capex_usd":4500,"power_tdp_watts":575,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://huggingface.co/nvidia/Cosmos-Predict1-7B-Text2World","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2501.03575","code_url":"https://github.com/nvidia-cosmos/cosmos-predict1","project_url":"https://www.nvidia.com/en-us/ai/cosmos/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010567,"per_frame_max":0.022187,"per_action_step_min":0.003694,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":50,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/cosmos\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/cosmos-drive-dreams","name":"Cosmos-Drive-Dreams","title":"Cosmos-Drive-Dreams: Scalable Synthetic Driving Data Generation with World Foundation Models","domain":"driving","domains":["driving"],"family":"driving","families":["driving"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos-Transfer1-7B-Sample-AV","weights_gated":null,"license_class":"conditional","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":["bf16"],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["map","layout","language","lidar","video"],"modalities_out":["video","lidar"],"model_type":"driving-world-model","task_suitability":"driving","environment_type":"outdoor-road","real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"published","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights are public. (https://huggingface.co/nvidia/Cosmos-Transfer1-7B-Sample-AV)","vendor":null,"weights_url":"https://huggingface.co/nvidia/Cosmos-Transfer1-7B-Sample-AV","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","multicam"],"outputs":["video"]},"embodiments":["vehicle"],"action_spaces":["steer_accel_2d","waypoint_se2","trajectory_bev"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.09042","code_url":"https://github.com/nv-tlabs/Cosmos-Drive-Dreams","project_url":"https://research.nvidia.com/labs/toronto-ai/cosmos_drive_dreams","providers":["baseten","fal","sidekick-cloud","sidekick-edge"],"pricing":{"per_frame_min":0.014937,"per_frame_max":0.026343,"per_action_step_min":0.005228,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/cosmos-drive-dreams\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"steer_accel_2d\",\n)"}},{"id":"community/cosmos-predict-2-5-cosmos-transfer-2-5","name":"Cosmos Predict 2.5 & Cosmos Transfer 2.5","title":"World Simulation with Video Foundation Models for Physical AI","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos-Predict2.5-2B","weights_gated":true,"license_class":"conditional","parameter_count_b":2.0,"vram_min_gb":4.8,"supported_quantizations":["bf16","fp32"],"vram_at_4bit_gb":2.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["language","rgb","video","action"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":null,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (nvidia/Cosmos-Predict2.5-2B, gated: accept terms first).","vendor":null,"weights_url":"https://huggingface.co/nvidia/Cosmos-Predict2.5-2B","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2511.00062","code_url":"https://github.com/nvidia-cosmos","project_url":"https://huggingface.co/blog/nvidia/cosmos-predict-and-transfer2-5","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010151,"per_frame_max":0.021181,"per_action_step_min":0.003549,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/cosmos-predict-2-5-cosmos-transfer-2-5\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/diffusion-forcing","name":"Diffusion Forcing","title":"Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/file/d/1xAOQxWcLzcFyD4zc0_rC9jGXe_uaHb7b/view?usp=sharing","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["video","trajectory"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"verified","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Inferred-tag check 2026-10-11: weights found (https://drive.google.com/file/d/1xAOQxWcLzcFyD4zc0_rC9jGXe_uaHb7b/view?usp=sharing).","vendor":null,"weights_url":"https://drive.google.com/file/d/1xAOQxWcLzcFyD4zc0_rC9jGXe_uaHb7b/view?usp=sharing","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["latent"]},"embodiments":["agnostic"],"action_spaces":[],"year":2024,"arxiv_url":"https://arxiv.org/abs/2407.01392","code_url":"https://github.com/buoyancy99/diffusion-forcing","project_url":"https://boyuan.space/diffusion-forcing","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009293,"per_frame_max":0.016876,"per_action_step_min":0.003252,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/diffusion-forcing\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/dino-foresight","name":"DINO-Foresight","title":"DINO-Foresight: Looking into the Future with DINO","domain":"general","domains":["general"],"family":"latent","families":["latent"],"subsection":"6. Latent Space World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Sta8is/DINO-Foresight","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["latent"],"model_type":"latent-world-model","task_suitability":"driving","environment_type":"outdoor-road","real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"verified","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Public trained weights exist. (https://huggingface.co/api/models/Sta8is/DINO-Foresight)","vendor":null,"weights_url":"https://huggingface.co/Sta8is/DINO-Foresight","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2412.11673","code_url":"https://github.com/Sta8is/DINO-Foresight","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008972,"per_frame_max":0.016277,"per_action_step_min":0.00314,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":51,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/dino-foresight\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"community/dino-wm","name":"DINO-WM","title":"DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning","domain":"general","domains":["general"],"family":"latent","families":["latent"],"subsection":"6. Latent Space World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://osf.io/bmw48/?view_only=a56a296ce3b24cceaf408383a175ce28","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action"],"modalities_out":["latent"],"model_type":"latent-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Public trained weights exist. (https://github.com/gaoyuezhou/dino_wm)","vendor":null,"weights_url":"https://osf.io/bmw48/?view_only=a56a296ce3b24cceaf408383a175ce28","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2411.04983","code_url":"https://github.com/gaoyuezhou/dino_wm","project_url":"https://dino-wm.github.io/","providers":["baseten","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008432,"per_frame_max":0.017246,"per_action_step_min":0.002951,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/dino-wm\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"community/emu-3-5","name":"Emu 3.5","title":"Emu3.5: Native Multimodal Models are World Learners","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"BAAI/Emu3.5","weights_gated":null,"license_class":"permissive","parameter_count_b":34.1,"vram_min_gb":81.8,"supported_quantizations":["bf16"],"vram_at_4bit_gb":34.1,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["image","text"],"model_type":"world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://huggingface.co/api/models/BAAI/Emu3.5)","vendor":null,"weights_url":"https://huggingface.co/BAAI/Emu3.5","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2510.26583","code_url":"https://github.com/baaivision/Emu3.5","project_url":"https://emu.world/pages/web/landingPage","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008732,"per_frame_max":0.017148,"per_action_step_min":0.003056,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":50,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/emu-3-5\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/fantasyworld","name":"FantasyWorld","title":"FantasyWorld: Geometry-Consistent World Modeling via Unified Video and 3D Prediction","domain":"general","domains":["general"],"family":"vision-3d","families":["vision-3d"],"subsection":"3. Building World Models from 3D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"acvlab/FantasyWorld-Wan2.1-I2V-14B-480P","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","camera-pose"],"modalities_out":["video","depth","3d"],"model_type":"3d-4d-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights and code are both released. (https://github.com/Fantasy-AMAP/fantasy-world)","vendor":null,"weights_url":"https://huggingface.co/acvlab/FantasyWorld-Wan2.1-I2V-14B-480P","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2509.21657","code_url":"https://github.com/Fantasy-AMAP/fantasy-world","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008584,"per_frame_max":0.017276,"per_action_step_min":0.003004,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/fantasyworld\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/geometry-forcing","name":"Geometry Forcing","title":"Geometry Forcing: Marrying Video Diffusion and 3D Representation for Consistent World Modeling","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Haoyuwu/GeometryForcing","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","camera-pose","action"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.36,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://github.com/CIntellifusion/GeometryForcing)","vendor":null,"weights_url":"https://huggingface.co/Haoyuwu/GeometryForcing","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video","latent"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2507.07982","code_url":"https://github.com/CIntellifusion/GeometryForcing","project_url":"https://GeometryForcing.github.io","providers":["baseten","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008208,"per_frame_max":0.017721,"per_action_step_min":0.002873,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":40,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/geometry-forcing\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/gigaworld-1","name":"GigaWorld-1","title":"GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation","domain":"embodied","domains":["embodied"],"family":"policy-eval","families":["policy-eval"],"subsection":"7. World Models for Policy evaluation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"open-gigaai/Giga-World-1","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Trained weights are downloadable. (https://huggingface.co/api/models/open-gigaai/Giga-World-1)","vendor":null,"weights_url":"https://huggingface.co/open-gigaai/Giga-World-1","routable":true,"featured":true,"contracts":["act","evaluate"],"modalities":{"inputs":["rgb"],"outputs":["action","score"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2607.02642","code_url":"https://github.com/open-gigaai/giga-world-1","project_url":"https://open-gigaai.github.io/giga-world-1/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.0087,"per_frame_max":0.017626,"per_action_step_min":0.003045,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":45,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/gigaworld-1\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/i-jepa","name":"I-JEPA","title":"Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture","domain":"general","domains":["general"],"family":"latent","families":["latent"],"subsection":"6. Latent Space World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"facebook/ijepa_vith14_1k","weights_gated":null,"license_class":"non-commercial","parameter_count_b":0.63,"vram_min_gb":1.5,"supported_quantizations":[],"vram_at_4bit_gb":0.6,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["latent"],"model_type":"latent-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Pretrained checkpoints are downloadable. (https://github.com/facebookresearch/ijepa)","vendor":null,"weights_url":"https://huggingface.co/facebook/ijepa_vith14_1k","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","proprio"],"outputs":["latent"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2023,"arxiv_url":"https://arxiv.org/abs/2301.08243","code_url":"https://github.com/facebookresearch/ijepa","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008218,"per_frame_max":0.016331,"per_action_step_min":0.002876,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":46,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/i-jepa\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"community/kairos","name":"Kairos","title":"Kairos: A Regret-Aware Native World-Action Model Stack for Physical AI","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"ACERobotics/kairos-4B-robot-RoboTwin2.0","weights_gated":null,"license_class":"permissive","parameter_count_b":4.0,"vram_min_gb":9.6,"supported_quantizations":[],"vram_at_4bit_gb":4.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.62,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights and code are released. (https://github.com/kairos-agi/kairos)","vendor":null,"weights_url":"https://huggingface.co/ACERobotics/kairos-4B-robot-RoboTwin2.0","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.16533","code_url":"https://github.com/kairos-agi/kairos","project_url":"https://huggingface.co/papers/2606.16533","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.011117,"per_frame_max":0.022392,"per_action_step_min":0.003887,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":51,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/kairos\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/matrix-3d","name":"Matrix-3D","title":"Matrix-3D: Omnidirectional Explorable 3D World Generation","domain":"game","domains":["game"],"family":"game-sim","families":["game-sim"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Skywork/Matrix-3D","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":16.0,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","camera-pose"],"modalities_out":["video","3d"],"model_type":"3d-4d-world-model","task_suitability":"simulation","environment_type":"simulation-only","real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights are public. (https://huggingface.co/Skywork/Matrix-3D)","vendor":null,"weights_url":"https://huggingface.co/Skywork/Matrix-3D","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["virtual-agent"],"action_spaces":["discrete_button","keyboard_mouse","continuous_2d"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2508.08086","code_url":"https://github.com/SkyworkAI/Matrix-3D","project_url":"https://matrix-3d.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008185,"per_frame_max":0.015622,"per_action_step_min":0.002865,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":46,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/matrix-3d\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"discrete_button\",\n)"}},{"id":"community/matrix-game-2-0","name":"Matrix-Game 2.0","title":"Matrix-Game 2.0: An Open-Source, Real-Time, and Streaming Interactive World Model","domain":"game","domains":["game"],"family":"game-sim","families":["game-sim"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Skywork/Matrix-Game-2.0","weights_gated":null,"license_class":"permissive","parameter_count_b":1.8,"vram_min_gb":24.0,"supported_quantizations":[],"vram_at_4bit_gb":1.8,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-workstation","hardware_rank":3,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 desktop (32GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action"],"modalities_out":["video"],"model_type":"interactive-game-world-model","task_suitability":"game","environment_type":"simulation-only","real_robot_evaluated":null,"hardware_capex_usd":4500,"power_tdp_watts":575,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://huggingface.co/Skywork/Matrix-Game-2.0)","vendor":null,"weights_url":"https://huggingface.co/Skywork/Matrix-Game-2.0","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["virtual-agent"],"action_spaces":["discrete_button","keyboard_mouse","continuous_2d"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2508.13009","code_url":"https://github.com/SkyworkAI/Matrix-Game","project_url":"https://matrix-game-v2.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009334,"per_frame_max":0.017498,"per_action_step_min":0.003267,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/matrix-game-2-0\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"discrete_button\",\n)"}},{"id":"community/mimic-video","name":"mimic-video","title":"mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAs","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"jonpai/mimic-video","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.38,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Inferred-tag check 2026-10-11: weights found (jonpai/mimic-video).","vendor":null,"weights_url":"https://huggingface.co/jonpai/mimic-video","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2512.15692","code_url":"https://huggingface.co/jonpai/mimic-video","project_url":"https://mimic-video.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008261,"per_frame_max":0.015922,"per_action_step_min":0.002891,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/mimic-video\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/mindjourney","name":"MindJourney","title":"MindJourney: Test-Time Scaling with World Models for Spatial Reasoning","domain":"embodied","domains":["embodied"],"family":"navigation","families":["navigation"],"subsection":"3. World Models for Navigation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"yyuncong/MindJourney-World-Model","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","camera-pose"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Trained world-model weights are publicly downloadable. (https://huggingface.co/api/models/yyuncong/MindJourney-World-Model)","vendor":null,"weights_url":"https://huggingface.co/yyuncong/MindJourney-World-Model","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["mobile-base"],"action_spaces":["base_twist_2d","waypoint_se2"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2507.12508","code_url":"https://github.com/UMass-Embodied-AGI/MindJourney","project_url":"https://umass-embodied-agi.github.io/MindJourney","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009107,"per_frame_max":0.016017,"per_action_step_min":0.003188,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":46,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/mindjourney\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"base_twist_2d\",\n)"}},{"id":"community/mosim","name":"MoSim","title":"Neural Motion Simulator Pushing the Limit of World Models in Reinforcement Learning","domain":"general","domains":["general"],"family":"mbrl","families":["mbrl"],"subsection":"8. World Models in the context of Model-Based RL","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"wujiss1/MoSim_checkpoints","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","tactile"],"modalities_out":["video","action"],"model_type":"model-based-rl","task_suitability":null,"environment_type":"simulation-only","real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.38,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights publicly downloadable. (https://huggingface.co/api/models/wujiss1/MoSim_checkpoints)","vendor":null,"weights_url":"https://huggingface.co/wujiss1/MoSim_checkpoints","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","tactile"],"outputs":["video","action"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2504.07095","code_url":"https://github.com/OamIcs/MoSim","project_url":null,"providers":["fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008336,"per_frame_max":0.017351,"per_action_step_min":0.002918,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":43,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/mosim\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"community/motus","name":"Motus","title":"Motus: A Unified Latent Action World Model","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"motus-robotics/Motus","weights_gated":null,"license_class":"permissive","parameter_count_b":8.0,"vram_min_gb":24.0,"supported_quantizations":[],"vram_at_4bit_gb":8.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":30,"replan_hz":null,"action_chunk_len":48,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-workstation","hardware_rank":3,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 desktop (32GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["action","latent"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":4500,"power_tdp_watts":575,"tag_completeness":0.77,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Trained weights are public. (https://huggingface.co/api/models/motus-robotics/Motus)","vendor":null,"weights_url":"https://huggingface.co/motus-robotics/Motus","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb"],"outputs":["action","latent"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2512.13030","code_url":"https://github.com/thu-ml/Motus","project_url":"https://motus-robotics.github.io/motus","providers":["baseten","fal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008713,"per_frame_max":0.01737,"per_action_step_min":0.003049,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":40,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/motus\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/nwm","name":"NWM","title":"Navigation World Models","domain":"embodied","domains":["embodied"],"family":"navigation","families":["navigation"],"subsection":"3. World Models for Navigation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"facebook/nwm","weights_gated":null,"license_class":"non-commercial","parameter_count_b":1.0,"vram_min_gb":2.4,"supported_quantizations":[],"vram_at_4bit_gb":1.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"navigation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://huggingface.co/api/models/facebook/nwm)","vendor":null,"weights_url":"https://huggingface.co/facebook/nwm","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["mobile-base"],"action_spaces":["base_twist_2d","waypoint_se2"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2412.03572","code_url":"https://github.com/facebookresearch/nwm","project_url":"https://www.amirbar.net/nwm/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008685,"per_frame_max":0.017819,"per_action_step_min":0.00304,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":46,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/nwm\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"base_twist_2d\",\n)"}},{"id":"community/osvi-wm","name":"OSVI-WM","title":"OSVI-WM: One-Shot Visual Imitation for Unseen Tasks using World-Model-Guided Trajectory Generation","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/drive/folders/1ekflx2fGqf8gVPISV1Jckc77SbhsXCRc?usp=sharing","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["video","rgb"],"modalities_out":["trajectory"],"model_type":"visuomotor-policy","task_suitability":null,"environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.36,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://raw.githubusercontent.com/raktimgg/osvi-wm/main/README.md)","vendor":null,"weights_url":"https://drive.google.com/drive/folders/1ekflx2fGqf8gVPISV1Jckc77SbhsXCRc?usp=sharing","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2505.20425","code_url":"https://github.com/raktimgg/osvi-wm","project_url":null,"providers":["fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009128,"per_frame_max":0.017419,"per_action_step_min":0.003195,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":46,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/osvi-wm\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/pointworld","name":"PointWorld","title":"PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulation","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/PointWorld_models","weights_gated":null,"license_class":"conditional","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","depth","action"],"modalities_out":["3d"],"model_type":"3d-4d-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Public weights and code exist. (https://huggingface.co/api/models/nvidia/PointWorld_models)","vendor":null,"weights_url":"https://huggingface.co/nvidia/PointWorld_models","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2601.03782","code_url":"https://github.com/NVlabs/PointWorld","project_url":"https://point-world.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008854,"per_frame_max":0.015125,"per_action_step_min":0.003099,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/pointworld\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/slotformer","name":"SlotFormer","title":"SlotFormer: Unsupervised Visual Dynamics Simulation with Object-Centric Models","domain":"general","domains":["general"],"family":"object-centric","families":["object-centric"],"subsection":"7. Building World Models from an Object-Centric Perspective","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/drive/folders/15y21miKZsAVHOSQEZLbUBWRrsZzcd5QW","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"verified","modalities_out":"verified","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights and code exist. (https://github.com/Wuziyi616/SlotFormer/blob/master/docs/benchmark.md)","vendor":null,"weights_url":"https://drive.google.com/drive/folders/15y21miKZsAVHOSQEZLbUBWRrsZzcd5QW","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2022,"arxiv_url":"https://arxiv.org/abs/2210.05861","code_url":"https://github.com/pairlab/SlotFormer","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008456,"per_frame_max":0.016435,"per_action_step_min":0.00296,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":38,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/slotformer\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/univla","name":"UniVLA","title":"UniVLA: Unified Vision-Language-Action Model","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Yuqi1997/UniVLA","weights_gated":null,"license_class":"permissive","parameter_count_b":8.5,"vram_min_gb":20.4,"supported_quantizations":[],"vram_at_4bit_gb":8.5,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":10,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action","video"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.69,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Public trained weights exist. (https://huggingface.co/api/models/Yuqi1997/UniVLA)","vendor":null,"weights_url":"https://huggingface.co/Yuqi1997/UniVLA","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","text"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.19850","code_url":"https://github.com/baaivision/UniVLA","project_url":"https://robertwyq.github.io/univla.github.","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge"],"pricing":{"per_frame_min":0.008891,"per_frame_max":0.015992,"per_action_step_min":0.003112,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/univla\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/uwm","name":"UWM","title":"Unified World Models: Coupling Video and Action Diffusion for Pretraining on Large Robotic Datasets","domain":"embodied","domains":["embodied"],"family":"policy-learning","families":["policy-learning"],"subsection":"6. World Models x Policy Learning","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/drive/folders/1M4AuVLMRpSwOf_YAp56bV9AqyZI9ul6g?usp=sharing","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":10,"replan_hz":null,"action_chunk_len":16,"action_horizon":8,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.54,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Public trained weights exist. (https://github.com/WEIRDLabUW/unified-world-model)","vendor":null,"weights_url":"https://drive.google.com/drive/folders/1M4AuVLMRpSwOf_YAp56bV9AqyZI9ul6g?usp=sharing","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2504.02792","code_url":"https://github.com/WEIRDLabUW/unified-world-model","project_url":"https://weirdlabuw.github.io/uwm/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008616,"per_frame_max":0.014928,"per_action_step_min":0.003016,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":43,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/uwm\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/vid2world","name":"Vid2World","title":"Vid2World: Crafting Video Diffusion Models to Interactive World Models","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"thuml/Vid2World-RT1","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.45,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://huggingface.co/api/models/thuml/Vid2World-RT1)","vendor":null,"weights_url":"https://huggingface.co/thuml/Vid2World-RT1","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2505.14357","code_url":"https://github.com/thuml/Vid2World","project_url":"http://knightnemo.github.io/vid2world/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008579,"per_frame_max":0.01791,"per_action_step_min":0.003003,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":54,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/vid2world\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"community/vidar-2507-12898","name":"Vidar","title":"Vidar: Embodied Video Diffusion Model for Generalist Manipulation","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation","vla"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"yaofeng1998/Vidar-Pretrain-10000steps","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.36,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Downloadable weights exist. (https://huggingface.co/papers/2507.12898)","vendor":null,"weights_url":"https://huggingface.co/yaofeng1998/Vidar-Pretrain-10000steps","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2507.12898","code_url":"https://huggingface.co/yaofeng1998/Vidar-Pretrain-10000steps","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008609,"per_frame_max":0.016672,"per_action_step_min":0.003013,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/vidar-2507-12898\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/video-world-models-with-long-term-spatial-memory","name":"Video World Models with Long-term Spatial Memory","title":"Video World Models with Long-term Spatial Memory","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d","memory"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"ysmikey/spmem_ckpt","weights_gated":null,"license_class":"permissive","parameter_count_b":8.26,"vram_min_gb":19.8,"supported_quantizations":["bf16"],"vram_at_4bit_gb":8.3,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","camera-pose","language"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"verified","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Weights exist. (https://huggingface.co/api/models/ysmikey/spmem_ckpt)","vendor":null,"weights_url":"https://huggingface.co/ysmikey/spmem_ckpt","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.05284","code_url":"https://github.com/spmem/spmem","project_url":"https://spmem.github.io/","providers":["baseten","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008336,"per_frame_max":0.016916,"per_action_step_min":0.002918,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":43,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"community/video-world-models-with-long-term-spatial-memory\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"community/vpp","name":"VPP","title":"Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"yjguo/svd-robot","weights_gated":null,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":["fp32"],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":160.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":10,"action_horizon":10,"time_to_first_action_ms":160.0,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":165.2,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action"],"model_type":"world-action-model","task_suitability":null,"environment_type":null,"real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.62,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"published","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"unknown","environment_type":"unknown","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Trained weights are downloadable. (https://github.com/roboterax/video-prediction-policy)","vendor":null,"weights_url":"https://huggingface.co/yjguo/svd-robot","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action","latent"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2412.14803","code_url":"https://github.com/roboterax/video-prediction-policy","project_url":"https://video-prediction-policy.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008513,"per_frame_max":0.017391,"per_action_step_min":0.002979,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"community/vpp\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"community/worldgym","name":"WorldGym","title":"WorldGym: World Model as An Environment for Policy Evaluation","domain":"embodied","domains":["embodied"],"family":"policy-eval","families":["policy-eval"],"subsection":"7. World Models for Policy evaluation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/file/d/1uiRP2BuavapMsyP9Cbr25mi_ymk9SEJb/view?usp=sharing","weights_gated":null,"license_class":null,"parameter_count_b":0.609,"vram_min_gb":1.5,"supported_quantizations":[],"vram_at_4bit_gb":0.6,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb"],"modalities_out":["value"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Code and a downloadable trained checkpoint exist. (https://github.com/world-model-eval/world-model-eval)","vendor":null,"weights_url":"https://drive.google.com/file/d/1uiRP2BuavapMsyP9Cbr25mi_ymk9SEJb/view?usp=sharing","routable":true,"featured":true,"contracts":["evaluate"],"modalities":{"inputs":["rgb"],"outputs":["score"]},"embodiments":["arm"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.00613","code_url":"https://github.com/world-model-eval/world-model-eval","project_url":"https://world-model-eval.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008671,"per_frame_max":0.016593,"per_action_step_min":0.003035,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":56,"example":{"contract":"evaluate","endpoint":"/v1/evaluate","returns":"a job","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.evaluate(\n    model=\"community/worldgym\",\n    observations=[...],\n    scenario=\"...\",\n)"}},{"id":"damo/worldvla","name":"WorldVLA","title":"WorldVLA: Towards Autoregressive Action World Model","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Alibaba-DAMO-Academy/WorldVLA","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":5,"action_horizon":5,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","action"],"modalities_out":["action","video"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.62,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.21539","code_url":"https://github.com/alibaba-damo-academy/WorldVLA","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.00836,"per_frame_max":0.017769,"per_action_step_min":0.002926,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":40,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"damo/worldvla\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"dreamzero0/dreamzero","name":"DreamZero","title":"World Action Models are Zero-shot Policies","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"GEAR-Dreams/DreamZero-DROID","weights_gated":false,"license_class":"non-commercial","parameter_count_b":22.92,"vram_min_gb":55.0,"supported_quantizations":["bf16","nvfp4"],"vram_at_4bit_gb":22.9,"weights_precision":"bf16","quantized_builds":[{"precision":"nvfp4","rung":"4bit","hardware":["GB200"],"engine":"TensorRT","memory_gb":null,"kind":"shipped","source":"https://github.com/dreamzero0/dreamzero"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":3000.0,"latency_hardware":"H100","latency_gpu_count":2,"action_frequency_hz":15,"replan_hz":null,"action_chunk_len":24,"action_horizon":24,"time_to_first_action_ms":3000.0,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"2x H100 or GB200 (README: at least 2 GPUs)","engine":"dreamzero","min_gpus":2,"min_cuda_cap":"8.0","min_tflops_fp16":1978.8,"modalities_in":["multi-view-rgb","language","proprio"],"modalities_out":["action","video"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"general","real_robot_evaluated":true,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":1.0,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"published","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"published","engine":"published","min_gpus":"published","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (GEAR-Dreams/DreamZero-DROID).","vendor":null,"weights_url":"https://huggingface.co/GEAR-Dreams/DreamZero-DROID","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2602.15922","code_url":"https://github.com/dreamzero0/dreamzero","project_url":"https://dreamzero0.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010988,"per_frame_max":0.020917,"per_action_step_min":0.003842,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":45,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"dreamzero0/dreamzero\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"eloialonso/diamond","name":"DIAMOND","title":"Diffusion for World Modeling: Visual Details Matter in Atari","domain":"game","domains":["game"],"family":"game-sim","families":["game-sim"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"eloialonso/diamond","weights_gated":false,"license_class":"permissive","parameter_count_b":0.381,"vram_min_gb":0.9,"supported_quantizations":[],"vram_at_4bit_gb":0.4,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":100.0,"latency_hardware":"RTX 3090","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":100.0,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":71.0,"modalities_in":["rgb","action"],"modalities_out":["video","action","value"],"model_type":"model-based-rl","task_suitability":"game","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.77,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://huggingface.co/eloialonso/diamond","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["virtual-agent"],"action_spaces":["discrete_button","keyboard_mouse","continuous_2d"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2405.12399","code_url":"https://github.com/eloialonso/diamond","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008146,"per_frame_max":0.0169,"per_action_step_min":0.002851,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"eloialonso/diamond\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"discrete_button\",\n)"}},{"id":"gangweix/next-forcing","name":"Next Forcing","title":"Next Forcing: Causal World Modeling with Multi-Chunk Prediction","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"gangweix/next-forcing-base","weights_gated":false,"license_class":"permissive","parameter_count_b":5.09,"vram_min_gb":12.2,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.1,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["video","action","language"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"simulation-only","real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.77,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (gangweix/next-forcing-base).","vendor":null,"weights_url":"https://huggingface.co/gangweix/next-forcing-base","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.11187","code_url":"https://github.com/gangweix/next-forcing","project_url":"https://gangweix.github.io/next-forcing/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008774,"per_frame_max":0.017554,"per_action_step_min":0.003071,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":52,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"gangweix/next-forcing\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"ginwind/vla-jepa","name":"VLA-JEPA","title":"VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"ginwind/VLA-JEPA","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action","latent"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.54,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","text"],"outputs":["action","latent"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2602.10098","code_url":"https://github.com/ginwind/VLA-JEPA","project_url":"https://ginwind.github.io/VLA-JEPA/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008152,"per_frame_max":0.017844,"per_action_step_min":0.002853,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":55,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"ginwind/vla-jepa\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"guandeh17/self-forcing","name":"Self-Forcing","title":"Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"gdhe17/Self-Forcing","weights_gated":false,"license_class":"permissive","parameter_count_b":1.3,"vram_min_gb":24.0,"supported_quantizations":["fp8"],"vram_at_4bit_gb":1.3,"weights_precision":null,"quantized_builds":[{"precision":"fp8","rung":"fp8","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://github.com/guandeh17/Self-Forcing"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-workstation","hardware_rank":3,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 desktop (32GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["language"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":4500,"power_tdp_watts":575,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"derived","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (gdhe17/Self-Forcing).","vendor":null,"weights_url":"https://huggingface.co/gdhe17/Self-Forcing","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.08009","code_url":"https://github.com/guandeh17/Self-Forcing","project_url":"https://self-forcing.github.io/","providers":["baseten","fal","modal","sidekick-cloud","vendor-direct"],"pricing":{"per_frame_min":0.00812,"per_frame_max":0.011978,"per_action_step_min":0.002842,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":151,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"guandeh17/self-forcing\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"guhuangai/ladi-wm","name":"LaDi-WM","title":"LaDi-WM: A Latent Diffusion-based World Model for Predictive Manipulation","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/file/d/1E3X_RfdZISwOW2l5SWuMUSiJmRnX-0Dg/view","weights_gated":false,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["latent","action"],"model_type":"latent-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.46,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://drive.google.com/file/d/1E3X_RfdZISwOW2l5SWuMUSiJmRnX-0Dg/view","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["latent"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2505.11528","code_url":"https://github.com/GuHuangAI/LaDiWM","project_url":"https://guhuangai.github.io/LaDiWM.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008698,"per_frame_max":0.017716,"per_action_step_min":0.003044,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":46,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"guhuangai/ladi-wm\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"hoyyyaard/3dflowaction","name":"3DFlowAction","title":"3DFlowAction: Learning Cross-Embodiment Manipulation from 3D Flow World Model","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Hoyard/3DFlowAction","weights_gated":false,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","depth"],"modalities_out":["trajectory"],"model_type":"3d-4d-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.55,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://huggingface.co/Hoyard/3DFlowAction","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","pointcloud"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.06199","code_url":"https://github.com/Hoyyyaard/3DFlowAction/","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.012626,"per_frame_max":0.021314,"per_action_step_min":0.004419,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":43,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"hoyyyaard/3dflowaction\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"huggingface/smolvla","name":"SmolVLA","title":"SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"lerobot/smolvla_base","weights_gated":false,"license_class":"permissive","parameter_count_b":0.45,"vram_min_gb":2.0,"supported_quantizations":["bf16"],"vram_at_4bit_gb":0.5,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":30,"replan_hz":null,"action_chunk_len":50,"action_horizon":50,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":"lerobot","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"derived","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Small open checkpoint trained on community data, served by LeRobot.","vendor":"Hugging Face","weights_url":"https://huggingface.co/lerobot/smolvla_base","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","proprio","language"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.01844","code_url":"https://github.com/huggingface/lerobot","project_url":"https://huggingface.co/blog/smolvla","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008372,"per_frame_max":0.017796,"per_action_step_min":0.00293,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"huggingface/smolvla\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"jiangranlv/dywa","name":"DyWA","title":"DyWA: Dynamics-adaptive World Action Model for Generalizable Non-prehensile Manipulation","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Steve3zz/Dywa_abs_1view","weights_gated":false,"license_class":"non-commercial","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["depth","proprio"],"modalities_out":["action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.54,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2503.16806","code_url":"https://github.com/jiangranlv/DyWA","project_url":"https://pku-epic.github.io/DyWA/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008274,"per_frame_max":0.016553,"per_action_step_min":0.002896,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":45,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"jiangranlv/dywa\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"jiangranlv/lda-1b","name":"LDA-1B","title":"LDA-1B: Scaling Latent Dynamics Action Model via Universal Embodied Data Ingestion","domain":"embodied","domains":["embodied"],"family":"policy-learning","families":["policy-learning"],"subsection":"6. World Models x Policy Learning","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Wayer2/LDA-pretrain","weights_gated":false,"license_class":"non-commercial","parameter_count_b":1.6,"vram_min_gb":3.8,"supported_quantizations":[],"vram_at_4bit_gb":1.6,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":10,"replan_hz":null,"action_chunk_len":16,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["action","latent"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"human-adjacent","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["action","latent"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2602.12215","code_url":"https://github.com/jiangranlv/latent-dynamics-action","project_url":"https://pku-epic.github.io/LDA/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008555,"per_frame_max":0.017834,"per_action_step_min":0.002994,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"jiangranlv/lda-1b\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"kwsong0113/dfot","name":"DFoT","title":"History-Guided Video Diffusion","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d","memory"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"kiwhansong/DFoT","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["video","rgb"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (kiwhansong/DFoT).","vendor":null,"weights_url":"https://huggingface.co/kiwhansong/DFoT","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2502.06764","code_url":"https://github.com/kwsong0113/diffusion-forcing-transformer","project_url":"https://www.boyuan.space/history-guidance/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009134,"per_frame_max":0.015626,"per_action_step_min":0.003197,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"kwsong0113/dfot\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"largeworldmodel/lwm","name":"LWM","title":"World Model on Million-Length Video And Language With RingAttention","domain":"general","domains":["general"],"family":"language","families":["language"],"subsection":"4. Building World Models from Language Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"LargeWorldModel/LWM-1M-Jax","weights_gated":false,"license_class":"conditional","parameter_count_b":7.0,"vram_min_gb":16.8,"supported_quantizations":[],"vram_at_4bit_gb":7.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["language","rgb","video"],"modalities_out":["text","video"],"model_type":"language-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (LargeWorldModel/LWM-1M-Jax).","vendor":null,"weights_url":"https://huggingface.co/LargeWorldModel/LWM-1M-Jax","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb","text"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2024,"arxiv_url":"https://arxiv.org/abs/2402.08268","code_url":"https://github.com/LargeWorldModel/LWM","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","vendor-direct"],"pricing":{"per_frame_min":0.008169,"per_frame_max":0.011671,"per_action_step_min":0.002859,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":161,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"largeworldmodel/lwm\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"little-podi/adaworld","name":"AdaWorld","title":"AdaWorld: Learning Adaptable World Models with Latent Actions","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Little-Podi/AdaWorld","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["video","action"],"modalities_out":["video","latent"],"model_type":"video-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://huggingface.co/Little-Podi/AdaWorld","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["action","latent"]},"embodiments":["arm"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2503.18938","code_url":"https://github.com/Little-Podi/AdaWorld","project_url":"https://adaptable-world-model.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008709,"per_frame_max":0.0175,"per_action_step_min":0.003048,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"little-podi/adaworld\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"meta/cwm","name":"CWM","title":"CWM: An Open-Weights LLM for Research on Code Generation with World Models","domain":"general","domains":["general"],"family":"language","families":["language"],"subsection":"4. Building World Models from Language Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"facebook/cwm","weights_gated":true,"license_class":"non-commercial","parameter_count_b":32.58,"vram_min_gb":160.0,"supported_quantizations":["bf16"],"vram_at_4bit_gb":32.6,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"datacenter-gpu","hardware_rank":5,"deployment_target":"datacenter","recommended_hardware":"NVIDIA H200 (141GB) or B200 (180GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["language"],"modalities_out":["text"],"model_type":"language-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":40000,"power_tdp_watts":1000,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (facebook/cwm, gated: accept terms first).","vendor":null,"weights_url":"https://huggingface.co/facebook/cwm","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb","text"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2510.02387","code_url":"https://github.com/facebookresearch/cwm","project_url":"https://ai.meta.com/resources/models-and-libraries/cwm-downloads","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008951,"per_frame_max":0.01473,"per_action_step_min":0.003133,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"meta/cwm\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"meta/v-jepa","name":"V-JEPA","title":"Revisiting Feature Prediction for Learning Visual Representations from Video","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general","latent"],"subsection":"1. Foundation World Models","era":"classic","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://dl.fbaipublicfiles.com/jepa/vith16/vith16.pth.tar","weights_gated":false,"license_class":"non-commercial","parameter_count_b":0.63,"vram_min_gb":1.5,"supported_quantizations":[],"vram_at_4bit_gb":0.6,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["video"],"modalities_out":["latent"],"model_type":"latent-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://dl.fbaipublicfiles.com/jepa/vith16/vith16.pth.tar","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["proprio","rgb"],"outputs":["latent","video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2404.08471","code_url":"https://github.com/facebookresearch/jepa","project_url":"https://ai.meta.com/blog/v-jepa-yann-lecun-ai-model-video-joint-embedding-predictive-architecture/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010889,"per_frame_max":0.022344,"per_action_step_min":0.003807,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":48,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"meta/v-jepa\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"meta/v-jepa-2","name":"V-JEPA 2","title":"V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general","latent"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"facebook/vjepa2-vitg-fpc64-256","weights_gated":false,"license_class":"permissive","parameter_count_b":1.035,"vram_min_gb":2.5,"supported_quantizations":["fp32"],"vram_at_4bit_gb":1.0,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":16000.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":1,"action_horizon":1,"time_to_first_action_ms":16000.0,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":165.2,"modalities_in":["rgb","video","action","proprio"],"modalities_out":["latent","action"],"model_type":"latent-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"derived","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["latent","video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.09985","code_url":"https://github.com/facebookresearch/vjepa2","project_url":"https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.011015,"per_frame_max":0.022018,"per_action_step_min":0.003851,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"meta/v-jepa-2\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"meta/v-jepa-2-1","name":"V-JEPA 2.1","title":"V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://dl.fbaipublicfiles.com/vjepa2/vjepa2_1_vitG_384.pt","weights_gated":false,"license_class":"permissive","parameter_count_b":2.0,"vram_min_gb":4.8,"supported_quantizations":[],"vram_at_4bit_gb":2.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["video","rgb"],"modalities_out":["latent"],"model_type":"latent-world-model","task_suitability":"general-manipulation","environment_type":"general","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (https://dl.fbaipublicfiles.com/vjepa2/vjepa2_1_vitG_384.pt).","vendor":null,"weights_url":"https://dl.fbaipublicfiles.com/vjepa2/vjepa2_1_vitG_384.pt","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video","latent"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2603.14482","code_url":"https://github.com/facebookresearch/vjepa2","project_url":"https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge"],"pricing":{"per_frame_min":0.01074,"per_frame_max":0.018412,"per_action_step_min":0.003755,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":57,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"meta/v-jepa-2-1\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"microsoft/rho-base","name":"Rho Base","title":"Rho: A Foundation for Efficiently Adaptable VLA Models (base pretrained checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","foundation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"microsoft/rho-base","weights_gated":false,"license_class":"permissive","parameter_count_b":5.238,"vram_min_gb":12.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.2,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":50,"action_horizon":25,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open 5B checkpoints on Hugging Face under MIT, with a policy server in the rhobotics repo.","vendor":"Microsoft Research","weights_url":"https://huggingface.co/microsoft/rho-base","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","proprio","language"],"outputs":["action"]},"embodiments":["bimanual"],"action_spaces":["ee_dual_20"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2609.38164","code_url":"https://github.com/microsoft/rhobotics","project_url":"https://huggingface.co/collections/microsoft/rho","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.013782,"per_frame_max":0.027643,"per_action_step_min":0.004824,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":47,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"microsoft/rho-base\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_dual_20\",\n)"}},{"id":"microsoft/rho-fr3-duo","name":"Rho FR3 Duo","title":"Rho: A Foundation for Efficiently Adaptable VLA Models (Franka Robotics FR3 Duo checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"microsoft/rho-fr3-duo","weights_gated":false,"license_class":"permissive","parameter_count_b":5.238,"vram_min_gb":12.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.2,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":30,"replan_hz":null,"action_chunk_len":50,"action_horizon":25,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open 5B checkpoints on Hugging Face under MIT, with a policy server in the rhobotics repo.","vendor":"Microsoft Research","weights_url":"https://huggingface.co/microsoft/rho-fr3-duo","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","proprio","language"],"outputs":["action"]},"embodiments":["bimanual"],"action_spaces":["ee_dual_20"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2609.38164","code_url":"https://github.com/microsoft/rhobotics","project_url":"https://huggingface.co/collections/microsoft/rho","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.013808,"per_frame_max":0.028171,"per_action_step_min":0.004833,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":50,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"microsoft/rho-fr3-duo\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_dual_20\",\n)"}},{"id":"microsoft/rho-libero","name":"Rho LIBERO","title":"Rho: A Foundation for Efficiently Adaptable VLA Models (LIBERO benchmark checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"microsoft/rho-libero","weights_gated":false,"license_class":"permissive","parameter_count_b":5.248,"vram_min_gb":12.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.2,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":16,"action_horizon":8,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open 5B checkpoints on Hugging Face under MIT, with a policy server in the rhobotics repo.","vendor":"Microsoft Research","weights_url":"https://huggingface.co/microsoft/rho-libero","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","proprio","language"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2609.38164","code_url":"https://github.com/microsoft/rhobotics","project_url":"https://huggingface.co/collections/microsoft/rho","providers":["baseten","fal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.014785,"per_frame_max":0.027316,"per_action_step_min":0.005175,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":54,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"microsoft/rho-libero\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"microsoft/rho-roboeval","name":"Rho RoboEval","title":"Rho: A Foundation for Efficiently Adaptable VLA Models (RoboEval benchmark checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"microsoft/rho-roboeval","weights_gated":false,"license_class":"permissive","parameter_count_b":5.238,"vram_min_gb":12.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.2,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":32,"action_horizon":16,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open 5B checkpoints on Hugging Face under MIT, with a policy server in the rhobotics repo.","vendor":"Microsoft Research","weights_url":"https://huggingface.co/microsoft/rho-roboeval","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","proprio","language"],"outputs":["action"]},"embodiments":["bimanual"],"action_spaces":["ee_dual_20"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2609.38164","code_url":"https://github.com/microsoft/rhobotics","project_url":"https://huggingface.co/collections/microsoft/rho","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.014566,"per_frame_max":0.026698,"per_action_step_min":0.005098,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":49,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"microsoft/rho-roboeval\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_dual_20\",\n)"}},{"id":"microsoft/rho-ur-ai-trainer","name":"Rho UR AI Trainer","title":"Rho: A Foundation for Efficiently Adaptable VLA Models (Universal Robots AI Trainer checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"microsoft/rho-ur-ai-trainer","weights_gated":false,"license_class":"permissive","parameter_count_b":5.238,"vram_min_gb":12.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.2,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":18,"replan_hz":null,"action_chunk_len":50,"action_horizon":25,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open 5B checkpoints on Hugging Face under MIT, with a policy server in the rhobotics repo.","vendor":"Microsoft Research","weights_url":"https://huggingface.co/microsoft/rho-ur-ai-trainer","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","proprio","language"],"outputs":["action"]},"embodiments":["bimanual"],"action_spaces":["ee_dual_20"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2609.38164","code_url":"https://github.com/microsoft/rhobotics","project_url":"https://huggingface.co/collections/microsoft/rho","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.014216,"per_frame_max":0.027891,"per_action_step_min":0.004976,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"microsoft/rho-ur-ai-trainer\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_dual_20\",\n)"}},{"id":"microsoft/rho-yam-box","name":"Rho YAM Box","title":"Rho: A Foundation for Efficiently Adaptable VLA Models (i2RT Robotics YAM Box checkpoint)","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"microsoft/rho-yam-box","weights_gated":false,"license_class":"permissive","parameter_count_b":5.238,"vram_min_gb":12.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":5.2,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":30,"replan_hz":null,"action_chunk_len":50,"action_horizon":25,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open 5B checkpoints on Hugging Face under MIT, with a policy server in the rhobotics repo.","vendor":"Microsoft Research","weights_url":"https://huggingface.co/microsoft/rho-yam-box","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","multicam","proprio","language"],"outputs":["action"]},"embodiments":["bimanual"],"action_spaces":["ee_dual_20"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2609.38164","code_url":"https://github.com/microsoft/rhobotics","project_url":"https://huggingface.co/collections/microsoft/rho","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.014114,"per_frame_max":0.024381,"per_action_step_min":0.00494,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":50,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"microsoft/rho-yam-box\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_dual_20\",\n)"}},{"id":"nicklashansen/td-mpc2","name":"TD-MPC2","title":"TD-MPC2: Scalable, Robust World Models for Continuous Control","domain":"general","domains":["general"],"family":"mbrl","families":["mbrl"],"subsection":"8. World Models in the context of Model-Based RL","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nicklashansen/tdmpc2","weights_gated":false,"license_class":"permissive","parameter_count_b":0.317,"vram_min_gb":8.0,"supported_quantizations":[],"vram_at_4bit_gb":0.3,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["proprio","rgb"],"modalities_out":["action","latent","value"],"model_type":"model-based-rl","task_suitability":"general-manipulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.69,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Source list records no link, but code and 317-task checkpoints are public under MIT.","vendor":null,"weights_url":"https://huggingface.co/nicklashansen/tdmpc2","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["action"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2023,"arxiv_url":"https://arxiv.org/abs/2310.16828","code_url":"https://github.com/nicklashansen/tdmpc2","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.00835,"per_frame_max":0.016084,"per_action_step_min":0.002923,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"nicklashansen/td-mpc2\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"nv-tlabs/omnidreams","name":"OmniDreams","title":"NVIDIA OmniDreams: Real-Time Generative World Model for Closed-Loop Autonomous Vehicle Simulation","domain":"driving","domains":["driving"],"family":"driving","families":["driving"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/omni-dreams-models","weights_gated":true,"license_class":"conditional","parameter_count_b":2.0,"vram_min_gb":48.0,"supported_quantizations":[],"vram_at_4bit_gb":2.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":118.0,"latency_hardware":"GB300","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","map","trajectory"],"modalities_out":["video"],"model_type":"driving-world-model","task_suitability":"driving","environment_type":"outdoor-road","real_robot_evaluated":false,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (nvidia/omni-dreams-models, gated: accept terms first).","vendor":null,"weights_url":"https://huggingface.co/nvidia/omni-dreams-models","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","multicam"],"outputs":["video"]},"embodiments":["vehicle"],"action_spaces":["steer_accel_2d","waypoint_se2","trajectory_bev"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.03159","code_url":"https://github.com/nv-tlabs/omni-dreams","project_url":"https://research.nvidia.com/labs/sil/projects/omnidreams-blog/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.013689,"per_frame_max":0.027774,"per_action_step_min":0.004791,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"nv-tlabs/omnidreams\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"steer_accel_2d\",\n)"}},{"id":"nvidia/cosmos-3","name":"Cosmos 3","title":"Cosmos 3: Omnimodal World Models for Physical AI","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos3-Nano","weights_gated":false,"license_class":"permissive","parameter_count_b":15.75,"vram_min_gb":37.8,"supported_quantizations":["bf16","fp8"],"vram_at_4bit_gb":15.8,"weights_precision":"bf16","quantized_builds":[{"precision":"fp8","rung":"fp8","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://github.com/nvidia/cosmos/blob/main/inference_benchmarks.md"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":59830.0,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":59830.0,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":989.4,"modalities_in":["language","rgb","video","action","audio"],"modalities_out":["video","text","action","image"],"model_type":"world-action-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":true,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"derived","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (nvidia/Cosmos3-Nano). The family row: each checkpoint is served as its own model, nvidia/cosmos3-edge, nvidia/cosmos3-nano and nvidia/cosmos3-super.","vendor":null,"weights_url":"https://huggingface.co/nvidia/Cosmos3-Nano","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.02800","code_url":"https://github.com/nvidia/cosmos","project_url":"https://research.nvidia.com/labs/cosmos-lab/cosmos3","providers":["baseten","fal","modal","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010536,"per_frame_max":0.019905,"per_action_step_min":0.003683,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":53,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"nvidia/cosmos-3\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"nvidia/cosmos3-edge","name":"Cosmos3-Edge","title":"Cosmos 3: Omnimodal World Models for Physical AI (Cosmos3-Edge)","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos3-Edge","weights_gated":false,"license_class":"permissive","parameter_count_b":4.0,"vram_min_gb":9.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":4.0,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":1250.0,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":15,"replan_hz":0.94,"action_chunk_len":32,"action_horizon":16,"time_to_first_action_ms":1250.0,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":989.4,"modalities_in":["multi-view-rgb","language","proprio","action"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"derived","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"inferred","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"NVIDIA's 4B Cosmos3, a World Action Model (one Mixture-of-Transformers network that predicts video and actions together), released 20 July 2026 under OpenMDW-1.1 (\"ready for commercial and non-commercial use\"). Two offerings, two checkpoints of one model: /v1/predict runs the base nvidia/Cosmos3-Edge through vLLM-Omni (image-to-video, or forward dynamics from an action trajectory for one of its ten trained embodiments; deploy/modal/cosmos3_world.py, 1x H100), and /v1/act runs NVIDIA's DROID post-train of it, nvidia/Cosmos3-Edge-Policy-DROID, through NVIDIA's own policy service (deploy/modal/cosmos3_policy.py): three cameras, an 8-wide Franka state and a task in; 32 steps of seven joint positions and the gripper at 15 Hz out, of which NVIDIA's Edge blog executes 16 before replanning. NVIDIA's H100 figure is 1.25 s per policy call (PyTorch, [16, 8]). Not run by Sidekick on a GPU yet.","vendor":"NVIDIA","weights_url":"https://huggingface.co/nvidia/Cosmos3-Edge","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","multicam","text","proprio","action"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["joint_pos_7_grip"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.02800","code_url":"https://github.com/nvidia/cosmos","project_url":"https://research.nvidia.com/labs/cosmos-lab/cosmos3","providers":["baseten","fal","modal","modal-cosmos3-edge","modal-cosmos3-edge-act","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.035455,"per_action_step_min":0.002,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"nvidia/cosmos3-edge\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_7_grip\",\n)"}},{"id":"nvidia/cosmos3-nano","name":"Cosmos3-Nano","title":"Cosmos 3: Omnimodal World Models for Physical AI (Cosmos3-Nano)","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos3-Nano","weights_gated":false,"license_class":"permissive","parameter_count_b":15.75,"vram_min_gb":46.0,"supported_quantizations":["bf16"],"vram_at_4bit_gb":15.8,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":15,"replan_hz":0.47,"action_chunk_len":32,"action_horizon":32,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language","proprio","action"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"vendor-default","replan_hz":"derived","action_chunk_len":"published","action_horizon":"vendor-default","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"inferred","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"NVIDIA's 16B Cosmos3, a World Action Model, released 31 May 2026 under OpenMDW-1.1. /v1/predict runs the base nvidia/Cosmos3-Nano through vLLM-Omni (deploy/modal/cosmos3_world.py; transformer about 17 GiB in BF16, about 46 GiB peak for 720p video on one GPU), and /v1/act runs nvidia/Cosmos3-Nano-Policy-DROID, NVIDIA's \"open SOTA DROID robot policy\", through NVIDIA's own policy service (deploy/modal/cosmos3_policy.py): 32 steps of seven joint positions and the gripper at 15 Hz per call. NVIDIA publishes no single-GPU latency for the Nano policy. Not run by Sidekick on a GPU yet.","vendor":"NVIDIA","weights_url":"https://huggingface.co/nvidia/Cosmos3-Nano","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","multicam","text","proprio","action"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["joint_pos_7_grip"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.02800","code_url":"https://github.com/nvidia/cosmos","project_url":"https://research.nvidia.com/labs/cosmos-lab/cosmos3","providers":["baseten","fal","modal","modal-cosmos3-nano","modal-cosmos3-nano-act","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.033341,"per_action_step_min":0.002,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"nvidia/cosmos3-nano\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_7_grip\",\n)"}},{"id":"nvidia/cosmos3-super","name":"Cosmos3-Super","title":"Cosmos 3: Omnimodal World Models for Physical AI (Cosmos3-Super)","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos3-Super","weights_gated":false,"license_class":"permissive","parameter_count_b":64.0,"vram_min_gb":153.6,"supported_quantizations":["bf16"],"vram_at_4bit_gb":64.0,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"datacenter-gpu","hardware_rank":5,"deployment_target":"datacenter","recommended_hardware":"NVIDIA H200 (141GB) or B200 (180GB)","engine":null,"min_gpus":2,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","language","action"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":null,"real_robot_evaluated":null,"hardware_capex_usd":40000,"power_tdp_watts":1000,"tag_completeness":0.69,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"published","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"inferred","environment_type":"unknown","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"NVIDIA's 64B Cosmos3, a World Action Model, released 31 May 2026 under OpenMDW-1.1. /v1/predict only: vLLM-Omni on NVIDIA's stated minimum of two H200s (--cfg-parallel-size 2 --use-hsdp --hsdp-shard-size 2, about 61.5 GiB per GPU; deploy/modal/cosmos3_world.py). No act offering: NVIDIA has released DROID policy checkpoints for Edge and Nano only. About 55 s per 50-step video on 8x H200 (model card). Not run by Sidekick on a GPU yet.","vendor":"NVIDIA","weights_url":"https://huggingface.co/nvidia/Cosmos3-Super","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb","text","action"],"outputs":["video","action"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.02800","code_url":"https://github.com/nvidia/cosmos","project_url":"https://research.nvidia.com/labs/cosmos-lab/cosmos3","providers":["baseten","fal","modal","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010787,"per_frame_max":0.021457,"per_action_step_min":0.003771,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":49,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"nvidia/cosmos3-super\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"nvidia/dreamgen","name":"DreamGen","title":"DreamGen: Unlocking Generalization in Robot Learning through Video World Models","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos-Predict2-14B-Sample-GR00T-Dreams-GR1","weights_gated":null,"license_class":null,"parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["video","action"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.46,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Inferred-tag check 2026-10-11: weights found (nvidia/Cosmos-Predict2-14B-Sample-GR00T-Dreams-GR1).","vendor":null,"weights_url":"https://huggingface.co/nvidia/Cosmos-Predict2-14B-Sample-GR00T-Dreams-GR1","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2505.12705","code_url":"https://github.com/nvidia/GR00T-dreams","project_url":"https://research.nvidia.com/labs/gear/dreamgen/","providers":["baseten","fal","modal","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.00883,"per_frame_max":0.015149,"per_action_step_min":0.003091,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"nvidia/dreamgen\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"nvidia/gr00t-n1-7","name":"GR00T N1.7","title":"NVIDIA Isaac GR00T N1.7, a Foundation Model for Generalist Humanoid Robots","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/GR00T-N1.7-3B","weights_gated":false,"license_class":"conditional","parameter_count_b":3.455,"vram_min_gb":16.0,"supported_quantizations":["bf16"],"vram_at_4bit_gb":3.5,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":27.9,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":40,"action_horizon":40,"time_to_first_action_ms":27.9,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":"groot","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":989.4,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"general","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"derived","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open checkpoint plus a ZMQ PolicyServer in the Isaac-GR00T repo.","vendor":"NVIDIA","weights_url":"https://huggingface.co/nvidia/GR00T-N1.7-3B","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","proprio","language"],"outputs":["action"]},"embodiments":["humanoid","arm"],"action_spaces":["joint_pos_23","joint_pos_29","ee_delta_6d_grip"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2503.14734","code_url":"https://github.com/NVIDIA/Isaac-GR00T","project_url":"https://developer.nvidia.com/isaac/gr00t","providers":["baseten","fal","modal","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010596,"per_frame_max":0.019122,"per_action_step_min":0.003704,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"nvidia/gr00t-n1-7\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_23\",\n)"}},{"id":"nvidia/groot-dreams","name":"GROOT-Dreams","title":"Dream Come True — NVIDIA Isaac GR00T-Dreams Advances Robot Training With Synthetic Data and Neural Simulation","domain":"embodied","domains":["embodied"],"family":"locomotion","families":["locomotion"],"subsection":"4. World Models for Locomotion","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"nvidia/Cosmos-Predict2-14B-Sample-GR00T-Dreams-GR1","weights_gated":null,"license_class":"conditional","parameter_count_b":14.0,"vram_min_gb":56.4,"supported_quantizations":["bf16"],"vram_at_4bit_gb":14.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["video","trajectory"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"general","real_robot_evaluated":true,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Tag audit 2026-10-10: Follows from public code + downloadable GR00T-Dreams checkpoints. If the curator judges the Cosmos checkpoint to be a separate model, tier should still be 'integration' (official Apache-2.0 code), not (https://huggingface.co/api/models/nvidia/Cosmos-Predict2-14B-Sample-GR00T-Dreams-GR1)","vendor":"NVIDIA","weights_url":"https://huggingface.co/nvidia/Cosmos-Predict2-14B-Sample-GR00T-Dreams-GR1","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["humanoid"],"action_spaces":["joint_pos_23","joint_pos_29","ee_delta_6d_grip"],"year":2025,"arxiv_url":null,"code_url":"https://github.com/NVIDIA/GR00T-Dreams","project_url":"https://blogs.nvidia.com/blog/nvidia-gtc-washington-dc-2025-news/#gr00t-dreams","providers":["baseten","fal","modal","nvidia-nim","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.00877,"per_frame_max":0.015225,"per_action_step_min":0.003069,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":56,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"nvidia/groot-dreams\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_23\",\n)"}},{"id":"octo-models/octo-base","name":"Octo","title":"Octo: An Open-Source Generalist Robot Policy","domain":"embodied","domains":["embodied"],"family":"policy-learning","families":["policy-learning"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"rail-berkeley/octo-base-1.5","weights_gated":false,"license_class":"permissive","parameter_count_b":0.093,"vram_min_gb":0.2,"supported_quantizations":[],"vram_at_4bit_gb":0.1,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":77.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":4,"action_horizon":4,"time_to_first_action_ms":77.0,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":165.2,"modalities_in":["multi-view-rgb","language"],"modalities_out":["action"],"model_type":"robot-foundation-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"published","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Open transformer policy trained on Open X-Embodiment, MIT licensed.","vendor":"UC Berkeley / Stanford","weights_url":"https://huggingface.co/rail-berkeley/octo-base-1.5","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","proprio","language"],"outputs":["action"]},"embodiments":["arm","mobile-base"],"action_spaces":["ee_delta_6d_grip"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2405.12213","code_url":"https://github.com/octo-models/octo","project_url":"https://octo-models.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008732,"per_frame_max":0.017474,"per_action_step_min":0.003056,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":50,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"octo-models/octo-base\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"openhelix/ud-vla","name":"UD-VLA","title":"Unified Diffusion VLA: Vision-Language-Action Model via Joint Discrete Denoising Diffusion Process","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"chenpyyy/UD-VLA_CALVIN-ABCD","weights_gated":false,"license_class":"permissive","parameter_count_b":8.492,"vram_min_gb":20.4,"supported_quantizations":["bf16"],"vram_at_4bit_gb":8.5,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":10,"action_horizon":10,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["action","video"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","text","proprio"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2511.01718","code_url":"https://github.com/OpenHelix-Team/UD-VLA","project_url":"https://irpn-eai.github.io/UD-VLA.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008697,"per_frame_max":0.017904,"per_action_step_min":0.003044,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":56,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"openhelix/ud-vla\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"openvla/openvla-7b","name":"OpenVLA","title":"OpenVLA: An Open-Source Vision-Language-Action Model","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"openvla/openvla-7b","weights_gated":false,"license_class":"conditional","parameter_count_b":7.541,"vram_min_gb":18.1,"supported_quantizations":["bf16","int8","4bit"],"vram_at_4bit_gb":7.5,"weights_precision":"bf16","quantized_builds":[{"precision":"int8","rung":"int8","hardware":[],"engine":"bitsandbytes","memory_gb":10.2,"kind":"documented","source":"https://arxiv.org/pdf/2406.09246"},{"precision":"4bit","rung":"4bit","hardware":[],"engine":"bitsandbytes","memory_gb":7.0,"kind":"documented","source":"https://arxiv.org/pdf/2406.09246"}],"vram_at_4bit_verified":true,"cache_memory_mb":null,"inference_latency_ms":167.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":5,"replan_hz":null,"action_chunk_len":1,"action_horizon":1,"time_to_first_action_ms":167.0,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":"transformers","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":165.2,"modalities_in":["rgb","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":1.0,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"published","vram_at_4bit_verified":"published","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"published","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"7B checkpoint on Hugging Face under MIT, with published inference code.","vendor":"Stanford / Google / TRI","weights_url":"https://huggingface.co/openvla/openvla-7b","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","language"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2406.09246","code_url":"https://github.com/openvla/openvla","project_url":"https://openvla.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008882,"per_frame_max":0.01638,"per_action_step_min":0.003109,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"openvla/openvla-7b\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"orangesodahub/orv","name":"ORV","title":"ORV: 4D Occupancy-centric Robot Video Generation","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"gzzyyxy/orv-gen-model","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action","depth","language"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.64,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb","pointcloud"],"outputs":["video","occupancy"]},"embodiments":["arm"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.03079","code_url":"https://github.com/OrangeSodahub/ORV","project_url":"https://orangesodahub.github.io/ORV/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.012462,"per_frame_max":0.024894,"per_action_step_min":0.004362,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":40,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"orangesodahub/orv\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"perceptron/isaac-0-5","name":"Isaac 0.5","title":"Isaac 0.5: Percepts Scale Control","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation","foundation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"PerceptronAI/Isaac-0.5","weights_gated":false,"license_class":"permissive","parameter_count_b":35.724,"vram_min_gb":85.7,"supported_quantizations":["fp32","bf16"],"vram_at_4bit_gb":35.7,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":50,"action_horizon":50,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","video","language","proprio","action"],"modalities_out":["text","action","points","trajectory","value"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Apache-2.0 on both the weights and the code, verified from the LICENSE file in the HF repo and the model card's own metadata. 42 safetensors shards are public and ungated, and the deployable LeRobot package ships with them under lerobot_policy/ with its own config, pre and post processor pipelines and a pinned policy class (perceptron_isaac). That package is a LIBERO artifact: 7-d end effector actions, 8-d proprio, cameras image and wrist_image. Grounding is not served from it, so this model is registered for act only. Announced 2026-08-26.","vendor":"Perceptron AI","weights_url":"https://huggingface.co/PerceptronAI/Isaac-0.5","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","video","proprio","language"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["continuous_nd","ee_delta_6d_grip"],"year":2026,"arxiv_url":null,"code_url":"https://github.com/perceptron-ai-inc/isaac","project_url":"https://www.perceptron.inc","providers":["baseten","fal","modal","modal-sidekick-2","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.017458,"per_action_step_min":0.0028,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":55,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"perceptron/isaac-0-5\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"continuous_nd\",\n)"}},{"id":"physical-intelligence/pi-0","name":"π0","title":"A Vision-Language-Action Flow Model for General Robot Control","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"gs://openpi-assets/checkpoints/pi0_base","weights_gated":false,"license_class":"conditional","parameter_count_b":3.5,"vram_min_gb":8.4,"supported_quantizations":["bf16","fp32"],"vram_at_4bit_gb":3.5,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":73.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":50,"replan_hz":null,"action_chunk_len":50,"action_horizon":50,"time_to_first_action_ms":73.0,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":"openpi","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":165.2,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"human-adjacent","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":1.0,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"derived","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Weights and serving code released as openpi; runs behind the openpi websocket policy server.","vendor":"Physical Intelligence","weights_url":"https://huggingface.co/physical-intelligence/pi0-base","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","proprio","language"],"outputs":["action"]},"embodiments":["arm","bimanual"],"action_spaces":["ee_delta_6d_grip","joint_pos_7","joint_pos_14"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2410.24164","code_url":"https://github.com/Physical-Intelligence/openpi","project_url":"https://www.pi.website/blog/openpi","providers":["baseten","fal","modal","modal-sidekick-2","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.017272,"per_action_step_min":0.0012,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"physical-intelligence/pi-0\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"physical-intelligence/pi-0-5","name":"π0.5","title":"A Vision-Language-Action Model with Open-World Generalization","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"gs://openpi-assets/checkpoints/pi05_base","weights_gated":false,"license_class":"conditional","parameter_count_b":3.62,"vram_min_gb":8.7,"supported_quantizations":["bf16","fp32"],"vram_at_4bit_gb":3.6,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":76.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":50,"replan_hz":null,"action_chunk_len":50,"action_horizon":50,"time_to_first_action_ms":76.0,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 4090 (openpi: more than 8 GB for inference)","engine":"openpi","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":165.2,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action","text"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"human-adjacent","real_robot_evaluated":true,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":1.0,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"published","engine":"derived","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Released in openpi; knowledge-insulation upgrade over pi0.","vendor":"Physical Intelligence","weights_url":"https://storage.googleapis.com/openpi-assets/checkpoints/pi05_base","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","proprio","language"],"outputs":["action"]},"embodiments":["arm","mobile-base","bimanual"],"action_spaces":["ee_delta_6d_grip","joint_pos_7","joint_pos_14"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2504.16054","code_url":"https://github.com/Physical-Intelligence/openpi","project_url":"https://www.pi.website/blog/pi05","providers":["baseten","fal","modal","modal-sidekick-2","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.006,"per_frame_max":0.014989,"per_action_step_min":0.0012,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":56,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"physical-intelligence/pi-0-5\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"physical-intelligence/pi-0-fast","name":"π0-FAST","title":"FAST: Efficient Action Tokenization for Vision-Language-Action Models","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla","manipulation"],"subsection":null,"era":"current","kind":"policy","is_world_model":false,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"gs://openpi-assets/checkpoints/pi0_fast_base","weights_gated":false,"license_class":"conditional","parameter_count_b":2.92,"vram_min_gb":8.0,"supported_quantizations":["bf16"],"vram_at_4bit_gb":2.9,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":750.0,"latency_hardware":"RTX 4090","latency_gpu_count":1,"action_frequency_hz":50,"replan_hz":null,"action_chunk_len":null,"action_horizon":50,"time_to_first_action_ms":750.0,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":"openpi","min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":165.2,"modalities_in":["multi-view-rgb","proprio","language"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.92,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"published","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"derived","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"override","tier_reason":"Autoregressive variant of pi0 released in the same openpi repo.","vendor":"Physical Intelligence","weights_url":"https://storage.googleapis.com/openpi-assets/checkpoints/pi0_fast_base","routable":true,"featured":true,"contracts":["act"],"modalities":{"inputs":["rgb","proprio","language"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2501.09747","code_url":"https://github.com/Physical-Intelligence/openpi","project_url":"https://www.pi.website/research/fast","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.0086,"per_frame_max":0.016081,"per_action_step_min":0.00301,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":48,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"physical-intelligence/pi-0-fast\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"pku-yuangroup/helios","name":"Helios","title":"Helios: Real Real-Time Long Video Generation Model","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"BestWishYsh/Helios-Distilled","weights_gated":false,"license_class":"permissive","parameter_count_b":14.31,"vram_min_gb":34.3,"supported_quantizations":["fp32"],"vram_at_4bit_gb":14.3,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["language","rgb","video"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (BestWishYsh/Helios-Distilled).","vendor":null,"weights_url":"https://huggingface.co/BestWishYsh/Helios-Distilled","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2603.04379","code_url":"https://github.com/PKU-YuanGroup/Helios","project_url":"https://pku-yuangroup.github.io/Helios-Page/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008523,"per_frame_max":0.015348,"per_action_step_min":0.002983,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":49,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"pku-yuangroup/helios\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"qwen/qwen-agentworld","name":"Qwen-AgentWorld","title":"Qwen-AgentWorld: Language World Models for General Agents","domain":"general","domains":["general"],"family":"language","families":["language"],"subsection":"4. Building World Models from Language Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Qwen/Qwen-AgentWorld-35B-A3B","weights_gated":false,"license_class":"permissive","parameter_count_b":34.66,"vram_min_gb":83.2,"supported_quantizations":["bf16"],"vram_at_4bit_gb":34.7,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["language"],"modalities_out":["text"],"model_type":"language-world-model","task_suitability":"simulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (Qwen/Qwen-AgentWorld-35B-A3B).","vendor":null,"weights_url":"https://huggingface.co/Qwen/Qwen-AgentWorld-35B-A3B","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","text"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2606.24597","code_url":"https://github.com/QwenLM/Qwen-AgentWorld","project_url":"https://qwen.ai/blog?id=qwen-agentworld","providers":["fal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009525,"per_frame_max":0.016007,"per_action_step_min":0.003334,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":48,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"qwen/qwen-agentworld\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"robbyant/lingbot-va","name":"LingBot-VA","title":"Causal World Modeling for Robot Control","domain":"embodied","domains":["embodied"],"family":"policy-learning","families":["policy-learning"],"subsection":"6. World Models x Policy Learning","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"robbyant/lingbot-va-base","weights_gated":false,"license_class":"permissive","parameter_count_b":5.3,"vram_min_gb":24.0,"supported_quantizations":[],"vram_at_4bit_gb":5.3,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-workstation","hardware_rank":3,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 desktop (32GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language","action"],"modalities_out":["video","action"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":4500,"power_tdp_watts":575,"tag_completeness":0.69,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2601.21998","code_url":"https://github.com/Robbyant/lingbot-va","project_url":"https://technology.robbyant.com/lingbot-va","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008987,"per_frame_max":0.017517,"per_action_step_min":0.003145,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"robbyant/lingbot-va\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"robbyant/lingbot-world","name":"LingBot-World","title":"Advancing Open-source World Models","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"robbyant/lingbot-world-base-cam","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":["nf4"],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[{"precision":"nf4","rung":"4bit","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://github.com/Robbyant/lingbot-world"}],"vram_at_4bit_verified":true,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","action","camera-pose"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"published","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"derived","vram_at_4bit_verified":"published","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (robbyant/lingbot-world-base-cam).","vendor":null,"weights_url":"https://huggingface.co/robbyant/lingbot-world-base-cam","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2026,"arxiv_url":"https://arxiv.org/abs/2601.20540","code_url":"https://github.com/Robbyant/lingbot-world/","project_url":"https://technology.robbyant.com/lingbot-world","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010275,"per_frame_max":0.020774,"per_action_step_min":0.003592,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":43,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"robbyant/lingbot-world\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"robert-gyj/ctrl-world","name":"Ctrl-World","title":"Ctrl-World: A Controllable Generative World Model for Robot Manipulation","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"yjguo/Ctrl-World","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":5000.0,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":15,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":989.4,"modalities_in":["multi-view-rgb","action","proprio"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2510.10125","code_url":"https://github.com/Robert-gyj/Ctrl-World","project_url":"https://ctrl-world.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008696,"per_frame_max":0.01611,"per_action_step_min":0.003043,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"robert-gyj/ctrl-world\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"shuangli59/uva","name":"UVA","title":"UVA, Unified Video Action Model","domain":"embodied","domains":["embodied"],"family":"policy-learning","families":["policy-learning"],"subsection":"6. World Models x Policy Learning","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"https://drive.google.com/file/d/11c2VrmaRp48yw__5A5xpcu8EPzkexHSi/view","weights_gated":null,"license_class":"permissive","parameter_count_b":0.5,"vram_min_gb":1.2,"supported_quantizations":[],"vram_at_4bit_gb":0.5,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":95.0,"latency_hardware":"RTX 3080","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":16,"action_horizon":8,"time_to_first_action_ms":95.0,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action","language"],"modalities_out":["action","video"],"model_type":"world-action-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.85,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"unknown","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"published","action_horizon":"published","time_to_first_action_ms":"derived","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2503.00200","code_url":"https://github.com/ShuangLI59/unified_video_action","project_url":"https://unified-video-action-model.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008731,"per_frame_max":0.017858,"per_action_step_min":0.003056,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":41,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"shuangli59/uva\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"skyworkai/matrix-game","name":"Matrix-Game","title":"Matrix-Game: Interactive World Foundation Model","domain":"game","domains":["game"],"family":"game-sim","families":["game-sim"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Skywork/Matrix-Game","weights_gated":false,"license_class":"permissive","parameter_count_b":17.0,"vram_min_gb":80.0,"supported_quantizations":[],"vram_at_4bit_gb":17.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action"],"modalities_out":["video"],"model_type":"interactive-game-world-model","task_suitability":"game","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://huggingface.co/Skywork/Matrix-Game","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["virtual-agent"],"action_spaces":["discrete_button","keyboard_mouse","continuous_2d"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2506.18701","code_url":"https://github.com/SkyworkAI/Matrix-Game","project_url":null,"providers":["baseten","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008728,"per_frame_max":0.017904,"per_action_step_min":0.003055,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":51,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"skyworkai/matrix-game\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"discrete_button\",\n)"}},{"id":"skyworkai/matrix-game-3-0","name":"Matrix-Game 3.0","title":"Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memory","domain":"game","domains":["game"],"family":"game-sim","families":["game-sim"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"Skywork/Matrix-Game-3.0","weights_gated":false,"license_class":"permissive","parameter_count_b":5.0,"vram_min_gb":12.0,"supported_quantizations":["int8"],"vram_at_4bit_gb":5.0,"weights_precision":null,"quantized_builds":[{"precision":"int8","rung":"int8","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://raw.githubusercontent.com/SkyworkAI/Matrix-Game/main/Matrix-Game-3/README.md"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-gpu-laptop","hardware_rank":2,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 Laptop GPU (24GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","action"],"modalities_out":["video"],"model_type":"interactive-game-world-model","task_suitability":"game","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":3500,"power_tdp_watts":175,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"derived","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (Skywork/Matrix-Game-3.0).","vendor":null,"weights_url":"https://huggingface.co/Skywork/Matrix-Game-3.0","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["virtual-agent"],"action_spaces":["discrete_button","keyboard_mouse","continuous_2d"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2604.08995","code_url":"https://github.com/SkyworkAI/Matrix-Game/tree/main/Matrix-Game-3","project_url":"https://matrix-game-v3.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008889,"per_frame_max":0.016993,"per_action_step_min":0.003111,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":38,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"skyworkai/matrix-game-3-0\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"discrete_button\",\n)"}},{"id":"tencent/hunyuanworld-1-0","name":"HunyuanWorld 1.0","title":"HunyuanWorld 1.0: Generating Immersive, Explorable, and Interactive 3D Worlds from Words or Pixels","domain":"game","domains":["game"],"family":"game-sim","families":["game-sim"],"subsection":null,"era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"tencent/HunyuanWorld-1","weights_gated":false,"license_class":"conditional","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":["fp8"],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[{"precision":"fp8","rung":"fp8","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://github.com/Tencent-Hunyuan/HunyuanWorld-1.0"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["language","rgb"],"modalities_out":["3d"],"model_type":"3d-4d-world-model","task_suitability":"simulation","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"published","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"derived","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (tencent/HunyuanWorld-1).","vendor":null,"weights_url":"https://huggingface.co/tencent/HunyuanWorld-1","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["virtual-agent"],"action_spaces":["discrete_button","keyboard_mouse","continuous_2d"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2507.21809","code_url":"https://github.com/Tencent-Hunyuan/HunyuanWorld-1.0","project_url":"https://3d-models.hunyuan.tencent.com/world/","providers":["sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.009714,"per_frame_max":0.017447,"per_action_step_min":0.0034,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":47,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"tencent/hunyuanworld-1-0\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"discrete_button\",\n)"}},{"id":"tencent/hy-world-1-5","name":"HY-World 1.5","title":"HY-World 1.5: A Systematic Framework for Interactive World Modeling with Real-Time Latency and Geometric Consistency","domain":"general","domains":["general"],"family":"foundation-general","families":["foundation-general"],"subsection":"1. Foundation World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"tencent/HY-WorldPlay","weights_gated":false,"license_class":"conditional","parameter_count_b":8.0,"vram_min_gb":72.0,"supported_quantizations":["fp8"],"vram_at_4bit_gb":8.0,"weights_precision":null,"quantized_builds":[{"precision":"fp8","rung":"fp8","hardware":[],"engine":null,"memory_gb":null,"kind":"listed","source":"https://github.com/Tencent-Hunyuan/HY-WorldPlay"}],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","action"],"modalities_out":["video"],"model_type":"interactive-game-world-model","task_suitability":"game","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"published","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"derived","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (tencent/HY-WorldPlay).","vendor":null,"weights_url":"https://huggingface.co/tencent/HY-WorldPlay","routable":true,"featured":true,"contracts":["predict","act","evaluate"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2512.14614","code_url":"https://github.com/Tencent-Hunyuan/HY-WorldPlay","project_url":"https://3d.hunyuan.tencent.com/login?redirect_url=https%3A%2F%2F3d.hunyuan.tencent.com%2FsceneTo3D","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge"],"pricing":{"per_frame_min":0.011154,"per_frame_max":0.019814,"per_action_step_min":0.0039,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":45,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"tencent/hy-world-1-5\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"thu-ml/causal-forcing","name":"Causal Forcing","title":"Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation","domain":"general","domains":["general"],"family":"vision-2d","families":["vision-2d"],"subsection":"2. Building World Models from 2D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"zhuhz22/Causal-Forcing","weights_gated":false,"license_class":"permissive","parameter_count_b":1.3,"vram_min_gb":3.1,"supported_quantizations":[],"vram_at_4bit_gb":1.3,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["language","rgb"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"n/a","environment_type":"general","real_robot_evaluated":false,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (zhuhz22/Causal-Forcing).","vendor":null,"weights_url":"https://huggingface.co/zhuhz22/Causal-Forcing","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":["latent_action","continuous_nd"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2602.02214","code_url":"https://github.com/thu-ml/Causal-Forcing","project_url":"https://thu-ml.github.io/CausalForcing.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008438,"per_frame_max":0.01656,"per_action_step_min":0.002953,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":38,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"thu-ml/causal-forcing\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"latent_action\",\n)"}},{"id":"thu-ml/ivideogpt","name":"iVideoGPT","title":"iVideoGPT: Interactive VideoGPTs are Scalable World Models","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"thuml/ivideogpt-oxe-64-act-free","weights_gated":false,"license_class":"permissive","parameter_count_b":0.252,"vram_min_gb":24.0,"supported_quantizations":[],"vram_at_4bit_gb":0.3,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"edge-workstation","hardware_rank":3,"deployment_target":"off-robot-lan","recommended_hardware":"RTX 5090 desktop (32GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","video","action"],"modalities_out":["video","value"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":false,"hardware_capex_usd":4500,"power_tdp_watts":575,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":"https://huggingface.co/thuml/ivideogpt-oxe-64-act-free","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2024,"arxiv_url":"https://arxiv.org/abs/2405.15223","code_url":"https://github.com/thuml/iVideoGPT","project_url":"https://thuml.github.io/iVideoGPT/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.011325,"per_frame_max":0.021119,"per_action_step_min":0.003959,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":38,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"thu-ml/ivideogpt\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"thu-ml/rlvr-world","name":"RLVR-World","title":"RLVR-World: Training World Models with Reinforcement Learning","domain":"general","domains":["general"],"family":"post-training","families":["post-training"],"subsection":"7. Post-training and Inference-Time Scaling for World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"thuml/rt1-world-model-multi-step-rlvr","weights_gated":false,"license_class":"permissive","parameter_count_b":0.127,"vram_min_gb":0.3,"supported_quantizations":["fp32"],"vram_at_4bit_gb":0.1,"weights_precision":"fp32","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","action","language"],"modalities_out":["video","text"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":null,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (thuml/rt1-world-model-multi-step-rlvr).","vendor":null,"weights_url":"https://huggingface.co/thuml/rt1-world-model-multi-step-rlvr","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb","tactile"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2505.13934","code_url":"https://github.com/thuml/RLVR-World","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008745,"per_frame_max":0.016235,"per_action_step_min":0.003061,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"thu-ml/rlvr-world\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"umass/tesseract","name":"TesserAct","title":"TesserAct: Learning 4D Embodied World Models","domain":"embodied","domains":["embodied"],"family":"manipulation","families":["manipulation"],"subsection":"2. World Models for Manipulation","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"anyeZHY/tesseract","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language"],"modalities_out":["video","3d","action"],"model_type":"3d-4d-world-model","task_suitability":"general-manipulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.54,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb","pointcloud"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2504.20995","code_url":"https://github.com/UMass-Embodied-AGI/TesserAct","project_url":"https://tesseractworld.github.io/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.011446,"per_frame_max":0.023218,"per_action_step_min":0.004006,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":43,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"umass/tesseract\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"world-grow/worldgrow","name":"WorldGrow","title":"WorldGrow: Generating Infinite 3D World","domain":"general","domains":["general"],"family":"vision-3d","families":["vision-3d"],"subsection":"3. Building World Models from 3D Vision Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"UranusITS/WorldGrow","weights_gated":false,"license_class":null,"parameter_count_b":null,"vram_min_gb":13.0,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":20000.0,"latency_hardware":"A100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":312.0,"modalities_in":["rgb"],"modalities_out":["3d"],"model_type":"3d-4d-world-model","task_suitability":"simulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.73,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"unknown","parameter_count_b":"unknown","vram_min_gb":"published","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"verified","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (UranusITS/WorldGrow).","vendor":null,"weights_url":"https://huggingface.co/UranusITS/WorldGrow","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2510.21682","code_url":"https://github.com/world-grow/WorldGrow","project_url":null,"providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.00884,"per_frame_max":0.017096,"per_action_step_min":0.003094,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":45,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"world-grow/worldgrow\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"wow-world-model/wow-2509-22642","name":"WoW","title":"WoW: Towards a World omniscient World model Through Embodied Interaction","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"WoW-world-model/WoW-1-Wan-14B-2M","weights_gated":false,"license_class":"permissive","parameter_count_b":14.0,"vram_min_gb":33.6,"supported_quantizations":[],"vram_at_4bit_gb":14.0,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["rgb","language","action"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (WoW-world-model/WoW-1-Wan-14B-2M).","vendor":null,"weights_url":"https://huggingface.co/WoW-world-model/WoW-1-Wan-14B-2M","routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2509.22642","code_url":"https://github.com/wow-world-model/wow-world-model","project_url":"https://wow-world-model.github.io","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010685,"per_frame_max":0.018909,"per_action_step_min":0.003736,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":42,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"wow-world-model/wow-2509-22642\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"xiaomi/xiaomi-robotics-u0","name":"Xiaomi-Robotics-U0","title":"Xiaomi-Robotics-U0: Unified Embodied Synthesis with World Foundation Model","domain":"embodied","domains":["embodied"],"family":"foundation","families":["foundation"],"subsection":"1. Foundation Embodied World Models","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"XiaomiRobotics/Xiaomi-Robotics-U0","weights_gated":false,"license_class":"permissive","parameter_count_b":34.1,"vram_min_gb":81.8,"supported_quantizations":["bf16"],"vram_at_4bit_gb":34.1,"weights_precision":"bf16","quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"pro-workstation","hardware_rank":4,"deployment_target":"off-robot-lan","recommended_hardware":"RTX PRO 6000 Blackwell (96GB)","engine":null,"min_gpus":null,"min_cuda_cap":"8.0","min_tflops_fp16":null,"modalities_in":["rgb","multi-view-rgb","depth","language"],"modalities_out":["image","video","text"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":12000,"power_tdp_watts":600,"tag_completeness":0.91,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"measured","vram_min_gb":"derived","supported_quantizations":"derived","vram_at_4bit_gb":"estimated","weights_precision":"published","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"derived","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2026,"arxiv_url":"https://arxiv.org/abs/2607.11643","code_url":"https://github.com/XiaomiRobotics/Xiaomi-Robotics-U0","project_url":"https://robotics.xiaomi.com/xiaomi-robotics-u0.html","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.010336,"per_frame_max":0.021964,"per_action_step_min":0.003613,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":40,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"xiaomi/xiaomi-robotics-u0\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"yuntian-group/neural-os","name":"Neural-OS","title":"NeuralOS: Towards Simulating Operating Systems via Neural Generative Models","domain":"general","domains":["general"],"family":"language","families":["language"],"subsection":"4. Building World Models from Language Priors","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"yuntian-deng/computer-model-s-newnewd-freezernn-origunet-nospatial-online-x0-joint-onlineonly-222222k7-136k","weights_gated":false,"license_class":null,"parameter_count_b":2.463,"vram_min_gb":5.9,"supported_quantizations":[],"vram_at_4bit_gb":2.5,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":55.6,"latency_hardware":"H100","latency_gpu_count":1,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"embedded-jetson","hardware_rank":1,"deployment_target":"onboard","recommended_hardware":"NVIDIA Jetson AGX Orin / Orin NX 16GB","engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":989.4,"modalities_in":["rgb","action"],"modalities_out":["video"],"model_type":"interactive-game-world-model","task_suitability":"simulation","environment_type":"simulation-only","real_robot_evaluated":false,"hardware_capex_usd":2000,"power_tdp_watts":60,"tag_completeness":0.82,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"unknown","parameter_count_b":"published","vram_min_gb":"derived","supported_quantizations":"unknown","vram_at_4bit_gb":"estimated","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"published","latency_hardware":"published","latency_gpu_count":"derived","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"estimated","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"estimated","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"estimated","power_tdp_watts":"estimated","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","replan_hz","time_to_first_action_ms"],"tier":"servable","tier_source":"override","tier_reason":"Weights found during model tagging research 2026-10 (yuntian-deng/computer-model-s-newnewd-freezernn-origunet-nospatial-online-x0-joint-onlineonly-222222k7-136k).","vendor":null,"weights_url":"https://huggingface.co/yuntian-deng/computer-model-s-newnewd-freezernn-origunet-nospatial-online-x0-joint-onlineonly-222222k7-136k","routable":true,"featured":true,"contracts":["predict"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["agnostic"],"action_spaces":[],"year":2025,"arxiv_url":"https://arxiv.org/abs/2507.08800","code_url":"https://github.com/yuntian-group/neural-os","project_url":"https://neural-os.com/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008793,"per_frame_max":0.017465,"per_action_step_min":0.003078,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":44,"example":{"contract":"predict","endpoint":"/v1/predict","returns":"frames","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.predict(\n    model=\"yuntian-group/neural-os\",\n    observations=[...],\n    horizon=16,\n)"}},{"id":"zhangwenyao1/dreamvla","name":"DreamVLA","title":"DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge","domain":"embodied","domains":["embodied"],"family":"vla","families":["vla"],"subsection":"5. World Models x VLAs","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":true,"weights_repo":"WenyaoZhang/DreamVLA","weights_gated":false,"license_class":"permissive","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"unsized","hardware_rank":7,"deployment_target":"unknown","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["multi-view-rgb","language","proprio"],"modalities_out":["action"],"model_type":"vla","task_suitability":"general-manipulation","environment_type":"lab-tabletop","real_robot_evaluated":true,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":0.54,"needs_verification":false},"tag_basis":{"weights_available":"published","weights_repo":"published","weights_gated":"published","license_class":"derived","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"derived","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"published","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":[],"tier":"servable","tier_source":"derived","tier_reason":null,"vendor":null,"weights_url":null,"routable":true,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb","text"],"outputs":["video","action"]},"embodiments":["arm"],"action_spaces":["ee_delta_6d_grip","joint_pos_7"],"year":2025,"arxiv_url":"https://arxiv.org/abs/2507.04447","code_url":"https://github.com/Zhangwenyao1/DreamVLA","project_url":"https://zhangwenyao1.github.io/DreamVLA/","providers":["baseten","fal","modal","sidekick-cloud","sidekick-edge","vendor-direct"],"pricing":{"per_frame_min":0.008217,"per_frame_max":0.017592,"per_action_step_min":0.002876,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":39,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"zhangwenyao1/dreamvla\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"ee_delta_6d_grip\",\n)"}},{"id":"1x/1x-world-model","name":"1X World Model","title":"1X World Model","domain":"embodied","domains":["embodied"],"family":"locomotion","families":["locomotion"],"subsection":"4. World Models for Locomotion","era":"current","kind":"world-model","is_world_model":true,"tags":{"schema":"sidekick.tags/1","weights_available":false,"weights_repo":null,"weights_gated":null,"license_class":"proprietary","parameter_count_b":null,"vram_min_gb":null,"supported_quantizations":[],"vram_at_4bit_gb":null,"weights_precision":null,"quantized_builds":[],"vram_at_4bit_verified":false,"cache_memory_mb":null,"inference_latency_ms":null,"latency_hardware":null,"latency_gpu_count":null,"action_frequency_hz":null,"replan_hz":null,"action_chunk_len":null,"action_horizon":null,"time_to_first_action_ms":null,"hardware_class":"no-weights","hardware_rank":9,"deployment_target":"not-deployable","recommended_hardware":null,"engine":null,"min_gpus":null,"min_cuda_cap":null,"min_tflops_fp16":null,"modalities_in":["video","action"],"modalities_out":["video"],"model_type":"video-world-model","task_suitability":"general-manipulation","environment_type":"human-adjacent","real_robot_evaluated":null,"hardware_capex_usd":null,"power_tdp_watts":null,"tag_completeness":1.0,"needs_verification":false},"tag_basis":{"weights_available":"verified","weights_repo":"unknown","weights_gated":"unknown","license_class":"verified","parameter_count_b":"unknown","vram_min_gb":"unknown","supported_quantizations":"unknown","vram_at_4bit_gb":"unknown","weights_precision":"unknown","quantized_builds":"unknown","vram_at_4bit_verified":"derived","cache_memory_mb":"unknown","inference_latency_ms":"unknown","latency_hardware":"unknown","latency_gpu_count":"unknown","action_frequency_hz":"unknown","replan_hz":"unknown","action_chunk_len":"unknown","action_horizon":"unknown","time_to_first_action_ms":"unknown","hardware_class":"verified","hardware_rank":"derived","deployment_target":"derived","recommended_hardware":"unknown","engine":"unknown","min_gpus":"unknown","min_cuda_cap":"unknown","min_tflops_fp16":"unknown","modalities_in":"published","modalities_out":"published","model_type":"verified","task_suitability":"verified","environment_type":"verified","real_robot_evaluated":"unknown","hardware_capex_usd":"unknown","power_tdp_watts":"unknown","tag_completeness":"derived"},"tag_na":["action_chunk_len","action_frequency_hz","action_horizon","cache_memory_mb","engine","hardware_capex_usd","min_cuda_cap","min_gpus","min_tflops_fp16","power_tdp_watts","recommended_hardware","replan_hz","supported_quantizations","time_to_first_action_ms","vram_at_4bit_gb","vram_min_gb","weights_gated","weights_precision","weights_repo"],"tier":"closed","tier_source":"override","tier_reason":"Humanoid world model, internal. Partner target for the humanoid embodiment.","vendor":"1X Technologies","weights_url":null,"routable":false,"featured":true,"contracts":["predict","act"],"modalities":{"inputs":["rgb"],"outputs":["video"]},"embodiments":["humanoid"],"action_spaces":["joint_pos_23","joint_pos_29","ee_delta_6d_grip"],"year":null,"arxiv_url":null,"code_url":null,"project_url":"https://www.1x.tech/discover/1x-world-model","providers":[],"pricing":{"per_frame_min":null,"per_frame_max":null,"per_action_step_min":null,"per_call_min":null,"basis":"units","rate_cards":null,"display":null,"indicative_call_usd_min":null,"indicative_assumes":null,"currency":"USD"},"latency_p50_ms_min":null,"example":{"contract":"act","endpoint":"/v1/act","returns":"an action chunk","python":"import os\nfrom sidekick_sdk import Sidekick\n\nsk = Sidekick(api_key=os.environ[\"SIDEKICK_API_KEY\"])\nresult = sk.act(\n    model=\"1x/1x-world-model\",\n    observations=[...],\n    instruction=\"fold the towel\",\n    action_space=\"joint_pos_23\",\n)"}}]}