AI MODELS
Follow important model changes, understand what changed, and check the official evidence.
New versions, capabilities, access and real-world deployments.
UnifoLM-WLA
Unitree released the Base checkpoint and fine-tuning code.
Previously reported
September 20: model modules and action-expert training code available
This announcement
September 28: Base weights and fine-tuning code available
UnifoLM-WLA
Unitree adds model modules and action-expert training code.
Previously reported
September 11: UnifoLM-ER-1 and ER-Flow weights released
This announcement
September 20: model modules and action-expert training code released
Figure Helix
Figure reports tests across 30 unseen homes.
Previously reported
Helix 02: data collected in deployment environments
This announcement
Helix 2.5: three whole-body tasks in 30 unseen homes
Search by model or company for versions, history and official resources.
Unitree Robotics
Vision-language-action
Unitree Robotics’ humanoid foundation model family for tabletop and whole-body manipulation.
Current versions: UnifoLM-WLA-1.0-Base
NVIDIA
NVIDIA Isaac GR00T is an open reference platform for general-purpose humanoid robots, combining open data pipelines, foundation models, simulation, middleware, accelerated runtime libraries, and Jetson Thor deployment.
4 Versions
Choose a version to compare
Figure AI
Figure AI vision-language-action system for humanoid robots.
Current versions: Helix 2.5
1X Technologies
1X world-model policy for the NEO humanoid robot.
1 Versions
UBTECH Robotics
UBTECH embodied-intelligence agent architecture for industrial humanoid robots.
1 Versions
Training-free VLM-based reward function for robot trajectories.
1 Versions
General-purpose vision-language reward model for robot task progress.
1 Versions
Stage-aware reward-modeling framework for long-horizon robot manipulation.
1 Versions
X Square Robot
Vision-language-action
Open vision-language-action family from X Square Robot.
1 Versions
Robot policy
Language-conditioned multitask diffusion-transformer robot policy.
1 Versions
Vision-language-action
Compact InternVL-based VLA family with flow-matching actions.
1 Versions
Robot policy · World model
World-action model family that learns future imagination but acts without rollout at inference.
1 Versions
Vision-language-action · World model
Autoregressive video-action world model family for robot control.
1 Versions
Vision-language-action
Interleaved vision-text-action pretraining family for general robot control.
1 Versions
Vision-language-action · World model
VLA family trained with an action-conditioned latent video world model.
1 Versions
Robot policy
Action Chunking with Transformers policy for fine-grained robot manipulation.
1 Versions
Vision-language-action
Ai2 action-reasoning model family for real-world robot deployment.
Current versions: MolmoAct2-Think / MolmoAct2
Choose a version to compare
Vision-language-action
Robotics Diffusion Transformer family for bimanual manipulation.
1 Versions
Vision-language-action
Open-source generalist robot policy family.
2 Versions
Choose a version to compare
Vision-language-action
Soft-prompted cross-embodiment VLA family.
1 Versions
Vision-language-action
Compact VLA family for accessible robot learning.
1 Versions
Vision-language-action
Physical Intelligence vision-language-action model family.
Current versions: π₀.₅
Choose a version to compare
NVIDIA
Vision-language-action · Robot policy
NVIDIA foundation model family for cross-embodiment robot control.
2 Versions
Choose a version to compare
Vision-language-action
Open vision-language-action model family for robot manipulation.
1 Versions
Unitree Robotics
Unitree’s Vision-Language-Action model family for humanoid manipulation.
1 Versions
