omnimodal/omnimodal-map

Omnimodal Map

An interactive architecture map for unified AI across text, vision, audio, video, documents, 3D, sensors, tools, memory, reasoning, world models, and actions.

What it covers

• —Input modalities
• —Shared representations
• —Cross-modal reasoning
• —Memory and context
• —World models
• —Tools and agents
• —Actions and physical AI
• —Validation and observability

Maintained by the Omnimodal organization on Hugging Face.

Research & industry collaborations: agenten@magenta.de