A free online demo now lets users type a text prompt and explore the interactive world that comes back, in real time.

California-based AI company Odyssey has opened a public research preview of its world model, Odyssey-3. The demo runs on Odyssey-3 Flash. You can move through the generated world in first-person or third-person view, trigger events and watch the model respond. Developers can apply for API access. Founders Oliver Cameron and Jeff Hawke first unveiled the model on September 15. What's new is public access, more technical detail and benchmark results.

Under the hood, Odyssey-3 is an autoregressive diffusion transformer. It keeps generating new video frames from previous frames and user actions. The base model has 14 billion parameters and outputs video at 832 × 480 pixels. Odyssey-3 Pro supports 1280 × 720.

On the Physics-IQ Verified video-to-video benchmark, Odyssey-3 Pro scored 66.1, according to the company. That came from a single run in which a selection method picked one of eight generated videos per task. The benchmark rules require four runs with standard deviation reported for any record claim, so it doesn't meet that bar. Without the selection method, the model averaged 63.37 across four runs. Odyssey submitted both results itself. On WorldMark, based on Odyssey's own evaluation, the model ranks first in three of four categories and third in First-Person Real.

Odyssey's goal is one world model driving robotic arms, drones and game characters, each paired with a specialized controller. The company showed an AI playing GTA V, and a controller trained on about two hours of GTA footage carried over to Red Dead Redemption 2. Robotics and autonomous systems need further work.

Would you spend an evening exploring a world you described in a single sentence?