Google beats Anthropic to decision models

🔥 What happened Google launched Gemini Decision 1, a model that makes closed-set decisions in a single pass instead of generating text token by token – beating Anthropic to production. It handles up to 1,000 questions per request at $0.10 per million input tokens with 150–250 ms latency. 💡 Why it matters Scoring choices directly is roughly 5x cheaper than Flash-Lite and returns a confidence score per answer. That's a new primitive for agents: Cloudflare's demo cut six LLM calls to one and 40K tokens to 1.8K by routing fast decisions to a small model. ⚡ Our take Google's six-week sprint from prototype to API is impressive, but independent evals are mixed – Decision 1 trails Drex Max by 7 points. The real test comes when developers rebuild their agent pipelines around it.