On Oct 8 Odyssey released Odyssey-3, a foundation world model built as an autoregressive diffusion transformer. Odyssey-3 Pro (720p) scores 66.1 on Physics-IQ Verified video-to-video, the highest reported score, and 54.7 image-to-video. A free browser research preview (Flash variant) offers first/third-person navigation and free camera; API access is by request, with no open weights or public pricing. Odyssey says tens of hours of demos are enough to control various robot arms.
Key Takeaways
- ✓Physics-IQ Verified v2v: Odyssey-3 Pro 66.1 (best-of-8, highest reported) / 63.4 prompt-enhanced; 480p model 51.8 → 61.6 → 64.4 (launch post)
- ✓Image-to-video: Odyssey-3 Pro 54.7; vendor says 1st in 3 of 4 WorldMark categories (self-reported)
- ✓Architecture: autoregressive diffusion transformer predicting environment evolution in real time from past observations plus latest actions/events
- ✓Physical AI: an action decoder trained on tens of hours of demos controls robot arms, with recovery behaviors absent from training data (technical write-up)
- ✓Access: free browser research preview; API by request, no open weights or public pricing
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Odyssey released Odyssey-3 on Oct 8, 2026, its most capable foundation world model. It is an autoregressive diffusion transformer that predicts in real time how an environment evolves given past observations and the latest actions or injected events. Two tiers ship: Odyssey-3 (480p) and Odyssey-3 Pro (720p); the free browser research preview runs a faster Flash variant with first/third-person navigation and independent camera.
On Physics-IQ Verified (Anates Labs and DeepMind; fluids, optics, solid mechanics, magnetism, thermodynamics), Odyssey-3 Pro reaches 66.1 video-to-video with best-of-8 (63.4 prompt-enhanced), the highest reported score; the 480p model scores 51.8 base, 61.6 prompt-enhanced and 64.4 best-of-8. Image-to-video: 54.7. Odyssey also claims 1st in 3 of 4 WorldMark categories in its own evaluations.
For physical AI, an action decoder trained on observation-action pairs maps the model's representations to controls; with tens of hours of demos it controls several robot arms and shows recovery behaviors absent from training data. Odyssey is benchmarking transfer with Poke & Wiggle. API access is by request; the public API docs still describe Odyssey-2 Pro. No open weights or pricing have been published.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.