Google has shipped two major AI releases at once: Gemini Omni and Gemini 3.5 Flash.

Gemini Omni is a new family of multimodal models that work across text, code, images, video, and even virtual worlds — not just understanding mixed inputs but editing, augmenting, and transforming them into new formats (e.g., adding effects to phone videos, turning rough sketches into finished artwork, or generating educational podcasts from photos). The first model in the family, Gemini Omni Flash, is available starting today.

Gemini 3.5 Flash is pitched as Google’s strongest model for agents and coding, with the ability to plan across large codebases, reason over long contexts, and spawn subagents in parallel. Google DeepMind claims it outperforms Gemini 3.1 Pro on agentic and coding benchmarks including Terminal-Bench 2.1, GDPval-AA, and MCP Atlas, while costing significantly less than frontier alternatives.

Google DeepMind announcement · Try it