Odyssey Trains World Model on Diablo II
Odyssey has introduced Agora-2, an AI model that generates the Diablo II game world in real-time without a traditional game engine.
The company Odyssey has trained a world model based on the cult game Diablo II.
There is no game engine under the hood at all. The AI generates every frame in real time.
Up to four humans and 16 agents can participate in a single session. Each has their own view of what is happening, but the world is shared: if someone moves or engages in combat, the consequences are visible to the others. Agents can already pursue players and navigate around obstacles.
In the future, the company plans to use such simulations to train agents on cooperation and competition with each other. For now, Agora-2 demonstrates this idea in a gaming environment.
Why it matters
AnalysisThis showcases progress in generative game engines and simulations for training multi-agent AI systems.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Thinking Machines Releases Inkling 2 Small: a 594B MoE Model That Can Fine-Tune Other Models
Thinking Machines released Inkling 2 Small, a 594B-parameter MoE model that leads among open models in several benchmarks and demonstrated autonomous fine-tuning of Qwen3.5-4B.
What If Claude Had Existed in 1998?
A short playful post about Anthropic’s Claude, imagining if the model had existed in 1998.
GPT Image Generative Models Prompting Guide
OpenAI's gpt-image generative models are designed to produce production-quality visuals and fully controllable creative workflows. They are suitable for both professional design ta...
Claude Sonnet 4.6 vs GPT-5.5 vs Gemini 3.1 Pro: 2026 model comparison
Every few weeks, a new model comes out, and every time the same questions arise: should I switch? what changed what to pay for?