Company
AI lab building world models, including the Odyssey-3 series, that generate interactive environments in real time and underpin policies for robots and driving.
Odyssey opened a free public research preview of Odyssey-3, a world model that generates interactive environments from text prompts in real time. The demo runs the Flash variant, while the 66.1 physics score Odyssey cites came from its Pro model on a best-of-eight run.
Reality
Mark de Kreij spotted verses from Book 12 on a scrap of parchment inside a collection Utrecht's library bought in the mid-20th century. About 30 comparable fragments are known, and his full catalogue is still to come.
Reality
Odyssey-3 keeps one visual world model frozen and trains small action decoders on top of it for arms, humanoids, drones, cars and game agents. Driving is the only test it ran in closed loop on public roads.
Reality
The capability claim is specific enough for a product team to test, with one image going in and up to a minute of 1440p video plus 3D splats coming out, though the camera-path win rests on a blind human study World Labs ran itself.
Reality
A new paper says Sora, Genie 3, JEPA and Marble carry no mental state at all. Its own benchmark, 420 of 448 scenes without motion, tests language models instead.
Reality
Techfundingnews frames the world model race as a $16B bet. The named rounds account for $2.31B of it, and the case against the architecture rests on a generalisation failure capital cannot fix.
Reality
The Cambridge company says routing agentic work across Nvidia, AMD and Google chips doubles accuracy at a quarter of the cost. No customer has confirmed those numbers.
Reality