Tuna-2

Tuna-2 is a native unified multimodal model developed by Meta that directly processes raw pixels without a VAE or vision encoder, using patch embeddings to achieve strong performance in both understanding and generation tasks.

Two-hop subgraph: this entity, every entity it directly relates to, and every entity those neighbors relate to. Drag a node, scroll to zoom, click to inspect — or click any neighbor and re-center the atlas there.