Odyssey-3 tests a video world model as a robot backbone
Odyssey-3 is a video world model that Odyssey says can also run robot arms, humanoids and cars. We check the claimed robot results against Physics-IQ Verified.
News and reference
World action models: robot policies that learn from video first.
A world action model starts from a model pretrained to predict video, then is fine-tuned to control a robot. We track the models, the papers, the benchmarks and the argument with vision-language-action models over which recipe wins.
- Pretrain on video. A backbone learns how scenes change from very large video corpora.
- Fine-tune on robot data. The same model learns to output actions.
- Imagine, then act. It predicts the next frames, the next actions, or both.
Top story
The stream
NVIDIA researcher maps world action models as robotics' second recipe
Moritz Reuss's long post sorts more than a dozen models into three paradigms, and predicts the field will end up with VLA–WAM hybrids.
The speed problem: why video-generating world action models are slower than VLAs
Imagining the future in pixels before acting costs hundreds of milliseconds. Representation-only designs try to keep the knowledge and skip the rendering.
What is a world action model?
A robot policy that starts from a model pretrained to predict video, then learns to act. Here is how it works, how it differs from a VLA, and what it costs.
Sections
- World action models
Policies that start from a video or world-model backbone and learn to predict both how a scene changes and what the robot should do.
- Vision-language-action models
The other recipe: policies built on vision-language models. Pi-0, GR00T and their successors, and the hybrids forming between the two.
- World models and video backbones
The pretrained models WAMs inherit from: Cosmos, Wan, V-JEPA, Genie and the video generators that learn how the world moves.
- Benchmarks and data
RoboArena, LIBERO, CALVIN, DROID and the other datasets and leaderboards that decide which recipe is winning.
I believe our generation was born too late to explore the Earth, and too early to explore the stars, but born just in time to solve robotics.