- Published on
Odyssey-3 is a new generative world model that you can try for free
- Authors
- Name
- DeepHorus
Odyssey, a California-based AI company, has launched a public research preview of its generative world model, Odyssey-3. The system generates interactive environments from text prompts and lets users explore them in real time. According to Odyssey, the model simulates physical processes and predicts how environments change based on specific actions. Developers can also apply for API access.
Founders Oliver Cameron and Jeff Hawke first unveiled the model on September 15, with an initial focus on robotics, autonomous driving, and video games. The new release adds public access, more technical detail, and benchmark results. The free online demo runs on Odyssey-3 Flash. Users can choose first-person or third-person perspectives, move through the generated world, trigger events, and watch the model respond in real time.
Odyssey-3 is built on an autoregressive diffusion transformer that continuously generates new video frames based on previous frames and user actions. The company says the model learns physical relationships and cause-and-effect from visual observations during training. Training data included internet videos with event descriptions, video game footage paired with corresponding keyboard and mouse inputs, and simulated physical interactions. An extra training technique reduces the number of required compute steps, making real-time generation possible.
The base Odyssey-3 model has 14 billion parameters and generates video at 832 Γ 480 pixels, according to the official benchmark submission. Odyssey-3 Pro supports 1280 Γ 720 pixels. The more powerful Pro model scores 66.1 points on the video-to-video benchmark from Physics-IQ Verified. That test evaluates physical behavior across fluid mechanics, optics, solid mechanics, magnetism, and thermodynamics. Models must continue videos of real-world experiments, and their outputs are compared against actual outcomes.
That top score comes with a major caveat. The 66.1 result is from a single test run where a selection method picked one of eight generated videos for each task. Benchmark rules require four test runs with standard deviation reported for any record claim. The reported record does not meet that bar. Without the selection method, Odyssey-3 Pro averaged 63.37 points across four runs. Both results appear on the official leaderboard but were submitted by Odyssey itself.
On the WorldMark benchmark, Odyssey-3 ranks first in three of four categories based on its own evaluation. It scores 77.2 in First-Person Stylized, 79.0 in Third-Person Real, and 76.3 in Third-Person Stylized. In First-Person Real, it places third with 80.6 points. WorldMark tests how well models follow control instructions, how good generated images look, and whether simulated worlds stay consistent over time.
Odyssey wants to use the same world model across different tasks, from moving robotic arms to navigating drones to controlling game characters. For each application, the model gets paired with a specialized controller that translates its predictions into concrete commands. The company positions Odyssey-3 as a foundation for controlling different systems, from robotic arms and drones in simulated environments to characters in video games.
The public research preview and API access are intended to let developers experiment with real-time interactive worlds. If the benchmark performance holds up under independent verification, Odyssey-3 could become a foundation for embodied AI and game development. However, the caveats around its top physics score highlight the need for standardized evaluation. Further results will show whether the model can deliver on its broader control ambitions.
