Research
Research notes
Technical notes and research writing from the Seele AI team on the foundation models and systems behind SeeleAgent — native 3D understanding and generation, multimodal architectures, and interactive world models. More notes are published here as our research progresses.
Notes & technical writing
EVA01 — native 3D understanding, generation, and multi-turn editing
How a Mixture-of-Transformers unifies 3D understanding, generation, and editing by treating mesh as a native language for multimodal models.
Read →PEGA-4world — real-time interactive world models
A physics-embedded generative architecture with long-term explicit 3D memory, built to simulate how game worlds evolve over time.
Read →PEGA — spatial simulation foundations
The Physics Embedded Generative Architecture direction powering spatial reasoning and world modeling in SeeleAgent.
Read →