Google launches the Gemini Decisions API for fast, low-cost yes/no, choice and rubric scoring
Google introduced the Gemini Decisions API on November 11, 2026, a Gemini API endpoint powered by the new gemini-decision-1 model that scores closed-set questions against text, image, audio, video or PDF input and returns probabilities and confidence scores rather than generated text.
What happened
- One request can check a shared multimodal input of up to 256K tokens against up to 1,000 yes/no, multiple-choice (2 to 255 options) or ordered-rubric questions, each answer with a 0-1 confidence score.
- Pricing is $0.10 per million input tokens with no output-token charges, and there is a no-cost tier for prototyping in Google AI Studio.
- Google reports 150-250 ms end-to-end latency and says it is nearly 5x cheaper on its benchmark suite than Gemini 3.5 Flash-Lite with the Interactions API.
- Google says the API already handles model routing in Gemini Notebook and a safety layer for MCP tools in Search, and that integrations are rolling out for OpenRouter, LangChain, Vercel, LiteLLM and others.
Why it matters: Apps and agents that make many small routing, tool-choice or policy decisions can hand them to a cheaper, faster model with calibrated confidence instead of a full generative call. Latency and cost comparisons are Google's own.
Sources
- Google: Gemini Decisions API: Fast, calibrated multimodal decisions blog.google
- Gemini API docs: Decisions API ai.google.dev
Checked November 12, 2026 against the sources above. Claims about performance or benchmarks are the company's own unless stated otherwise.