BagelVLA
Interleaves language planning, visual forecasting and action prediction.
RESEARCH DIRECTORY
Model directory
Explore 18 model profiles and their documented research connections. Editorial classifications do not establish comparable benchmarks or commercial availability.
Being-H0.7
Uses future-informed latent reasoning without generating future video frames at inference.
Cosmos Policy
Encodes robot actions as latent video frames while predicting future states and actions.
DreamDojo
Learns action-controllable robot-world simulations from large-scale human video.
DreamZero
Jointly predicts future visual states and robot actions using a pretrained video backbone.
DVA
Uses predicted motion and visual changes as part of robot-action generation.
Fast-WAM
Uses video-model representations to predict actions without generating full video at inference.
Genie
Generates controllable environments using video-trained latent actions.
GENIMA
Draws action targets into images and maps those targets to robot joint positions.
GR-1
Pretrains a video-predictive transformer before adapting it to language-conditioned robot actions.
LingBot-VA
A video-action model using joint video/action attention and a pretrained video backbone.
mimic-video
Decodes robot actions from the latent trajectory generated by a pretrained video model.
Motus
Combines understanding, video generation and action experts in one multi-purpose architecture.
Unified Video Action
Learns a joint representation for video and robot actions with separate lightweight decoders.
UniPi
An early world-action recipe: imagine a future video, then infer the actions to realise it.
V-JEPA 2
Learns predictive visual representations and supports action-conditioned robot planning.
Video Prediction Policy
Uses predictive visual representations from video diffusion models for robot manipulation.
WorldBridge
Links video-generalisation and robot action prediction through coordinated experts.