State-of-the-art coding, multi-step agent reasoning, and vision comprehension.

Flagship multimodal intelligence with high speed and broad domain versatility.

Ultra-cost-efficient MoE architecture rivaling top closed frontier models.

Open reasoning model with transparent chain-of-thought verification.

Sub-second multimodal model with 1M context window and cached input acceleration.

Google frontier model for deeply complex reasoning across massive multimodal contexts.

Fast, lightweight model for everyday classification, summaries, and agent sub-tasks.