Google launches Gemini Decision 1

Google's post on 11 November 2026 introduces the Gemini Decisions API, powered by Gemini Decision 1 (gemini-decision-1), through the Gemini API in Google AI Studio. One request carries a shared multimodal input, up to 256K tokens, across text, images, audio, video, and PDFs, plus up to 1,000 independent questions. The model scores the options in a single forward pass and returns probabilities and a confidence score. Google says end-to-end latency is 150–250 ms, at $0.10 per million input tokens, with no output tokens billed.

Three question types can sit in the same request. A noul estimates whether a yes/no condition holds. A choice selects from 2 to 255 options and returns the winner plus a probability for each. A score grades an ordered rubric and returns a weighted score. Confidence runs from 0 to 1. Google says a no-cost tier in Google AI Studio is there for prototyping, and that the shared input is billed once no matter how many questions ride along. Support in Gemini on Google Cloud is described as coming soon. The post lists ecosystem clients that are rolling out, including the google-genai SDKs.

Who it's for: a developer routing a request, checking a policy, or grading a rubric inside an app, where the result is a score rather than a paragraph.

Fits: Best for Writing stays Claude. Gemini stays third, for a draft next to Gmail and Docs. Decision 1 returns a probability, which leaves that draft alone. Best for Research stays Perplexity. The writing guide and the Gemini alternatives are unchanged. The verdict is unchanged.