Odyssey-2 is a general-purpose world model from Odyssey AI that simulates the world in video. It can generate minutes-long, interactive simulations from an image or text prompt, streaming frame-by-frame with a 50ms start time. It learns from decades of video data to simulate physics, dynamics, and human behaviors.

What is the difference between Odyssey-2 and Odyssey-2 Pro?

Odyssey-2 Pro is a significantly larger model with improved physics, better dynamics, more authentic behaviors, and more realistic pixels. It streams at 720P resolution at 22 FPS, compared to the base Odyssey-2. Odyssey-2 Pro is the first model available via the Odyssey API.

How do I get access to the Odyssey API?

You can get API access by visiting the Odyssey developer portal at odyssey.ml, generating an API key, and integrating the SDK (JavaScript or Python) into your application. The process takes under 10 lines of code and can be completed in 5 minutes or less.

What can I build with the Odyssey API?

The API enables building applications that require interactive, real-time world simulation: robotics rehearsal, immersive gaming, personalized education, healthcare navigation, intelligent customer support, virtual try-on, training simulations for defense, interactive advertising, and more. The model can be prompted with images or text to generate simulations.

Is there a free trial for Odyssey?

Yes, you can 'Try Odyssey-2 for free' on the Odyssey website. The app allows you to simulate your dreams in real-time without payment. Developers can also explore the API with a free tier (details not specified on the page).

Who are the investors behind Odyssey?

Odyssey has been backed by NVentures (NVIDIA's venture capital arm), Samsung Next, GV, EQT, Air Street Capital, Elad Gil, Jeff Dean, Guillermo Rauch, Garry Tan, and Kyle Vogt.

What team is behind Odyssey?

The leadership team includes Oliver Cameron (Co-Founder & CEO), Jeff Hawke (Co-Founder & CTO), James Grieve (VP Engineering), Jessica Inman (VP GTM & Operations), and Fabian Güra (Distinguished Engineer). The research team is recruited from DeepMind, OpenAI, ByteDance, Tesla, Waymo, Meta, Wayve, and Luma.

What SDKs does the Odyssey API support?

Currently, Odyssey supports JavaScript and Python SDKs. iOS and Android SDKs are coming soon.

What is the technical architecture of Odyssey-2?

Odyssey-2 uses Autoregressive Diffusion Transformers (AR DiTs), which generate frames sequentially, conditioning each new frame on all previous frames and action inputs. This causal, action-conditioned prediction model enables open-ended interaction and long-horizon stability.

Is Odyssey suitable for commercial use?

Yes, the Odyssey API is designed for commercial integration. The company encourages developers to build 'all kinds of new gaming, simulation, education, healthcare, companionship, support, training, and intelligence applications.' Specific licensing terms are not detailed on the page.

Odyssey: General-Purpose World Model API for Real-Time Interactive Simulation

Odyssey Overview

Odyssey is an AI research lab dedicated to building general-purpose world models — causal, multimodal systems that learn to predict and interact with the world over long horizons. Unlike traditional language models, Odyssey's models learn from trillions of video observations to simulate physics, dynamics, human behavior, and cause-and-effect relationships. The flagship product, Odyssey-2, is a nascent world simulator that can generate minutes-long, interactive video simulations from an image or text prompt in real-time (50ms start). The more advanced Odyssey-2 Pro streams at 720P 22 FPS with improved physics, richer interactivity, and authentic human behaviors. Odyssey offers a developer API with three endpoints — interactive streams, viewable streams, and simulations — along with JavaScript and Python SDKs to integrate world models into applications. The company is backed by NVentures (NVIDIA's VC arm), Samsung Next, GV, EQT, and other top investors.

How to use Odyssey

To use Odyssey, you can either try the free interactive demo on the website or integrate the API into your own application. For developers: sign up on the developer portal, generate an API key, and use the provided JavaScript or Python SDK to call endpoints such as client.startStream(), client.interact(), or client.simulate() in under 10 lines of code. You can provide an image or text prompt, and the model will begin streaming a simulation within 50ms. You can programmatically interact with the simulation in real-time (e.g., 'A kitten appears') or generate batch simulations with predefined actions.

Core Features of Odyssey

Real-time interactive world simulation: generates video frames sequentially in 50ms, allowing open-ended interaction
Minutes-long streaming: simulations can last minutes (not just 10-second clips)
Multi-modal input: accepts both images and text as prompts
Odyssey-2 Pro: streams 720P video at 22 FPS with enhanced physics, dynamics, and visual quality
Three API endpoints: interactive streams (real-time control), viewable streams (distribute to many users), simulations (batch with predefined actions)
Developer SDKs: JavaScript and Python SDKs (iOS and Android coming soon)
Causal, action-conditioned prediction: model predicts next state step-by-step based on current state and actions

Use Cases for Odyssey

Odyssey's world models unlock applications across robotics (simulating manipulation, navigation, and rehearsal before real actions), gaming (dynamic, interactive environments), education (personalized tutors, historical simulations), healthcare (navigation support, empathetic support), hospitality (intelligent guides), defense (realistic training scenarios), retail (instant try-on, interactive advertising), and entertainment (interactive art, immersive fitness). The API enables developers to build new types of software, hardware, and intelligence applications.

Advantages of Odyssey

Odyssey's world models go beyond language models by learning from the richest data source — video — which captures physics, human behavior, and causality at scale. The causal autoregressive diffusion transformer architecture enables stable, interactive long-horizon simulations that traditional video models cannot achieve. Instant streaming (50ms) eliminates waiting times. The model learns cause-and-effect relationships, making it suitable for robotics and scientific applications. Backed by world-class investors and a research team from DeepMind, OpenAI, Tesla, and Waymo, Odyssey is at the forefront of the world model paradigm shift.

Odyssey FAQ

Odyssey Comments (0)

Sign in to comment.

Sign in

No comments yet.

Traffic

Latest traffic

Monthly visits133.4K

Avg visit duration0:24

Pages per visit1.94

Bounce rate41.7%

Status

Rising+101.4%vs previous month

Updated at 2026-06-15

Monthly traffic trend

Geography

Top 5 countries / regions

🇺🇸United States
43.8%
🇮🇳India
24.1%
🇬🇧United Kingdom
12.9%
🇨🇭Switzerland
9.8%
🇩🇪Germany
9.4%

Traffic sources

Source type	Percentage
Direct	75.9%
Referral	21.2%
Email	2.8%

Total

100%

Direct75.9%

Referral21.2%

Email2.8%

Top keywords

Keyword	Cost per click
agora 1	$0.31
odyssey	$1.34
odyssey agora	$0.00
odyssey ai	$1.80
odyssey ml	$0.00

Odyssey Videos on YouTube