Models · Updated 11 Oct, 01:18 pm IST
Odyssey launches public preview of generative world model Odyssey-3
Why it matters for readers: It lets you create and play inside AI-generated interactive scenes from simple text prompts.
- Odyssey-3 generates interactive environments from text prompts and lets users explore them in first- or third-person in real time.1
- The model is built on an autoregressive diffusion transformer that produces new video frames conditioned on previous frames and user actions.1
- Odyssey says the model learned cause-and-effect and physical relationships from training on internet videos, video game footage with input traces, and simulated data.1
- Founders Oliver Cameron and Jeff Hawke unveiled Odyssey-3 on September 15 and the company released more technical detail, benchmarks, and public access now.1
- Developers can apply for API access, while anyone can try the free Odyssey-3 Flash demo online.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started