Glossary
World action model
A robot foundation model that predicts how the world will change and chooses actions from that prediction, rather than mapping images straight to motor commands.
Where a vision-language-action model outputs actions directly from what it sees and is told, a world action model first simulates the likely outcome of candidate actions using a learned world model, then acts. Nvidia's GR00T 2, built on its Cosmos world models, is the most prominent example; the approach is more compute-intensive per step but can generalise better to new tasks from human video.
Related terms
AgentAgent harnessBenchmarkNon-human identity
Get the brief in your inbox
One email every morning: the stories that mattered in AI, each with a one-line "why it matters" for your role. Free, no ads, unsubscribe any time.
Published daily by AI Briefing. We never share your address.