Google launches Gemini Decisions API and gemini-decision-1 model

Google introduced a new API and model built for calibrated yes/no, multiple-choice and rubric scoring instead of text generation, intended for agent routing, policy checks and tool filtering.

Google announced the Gemini Decisions API and a new gemini-decision-1 model on November 11, 2026, in a post on the Google blog. Unlike standard generative models, gemini-decision-1 is built specifically for calibrated yes/no, multiple-choice and rubric scoring, aimed at agent routing, policy checks and tool filtering inside agentic pipelines.

What changes with this release

New gemini-decision-1 model exposed through the Gemini Decisions API, scoring up to 1,000 independent closed-set questions in a single request.
Multimodal inputs supported: text, images, audio, video and PDFs.
Input context window up to 256K tokens, with up to 16K tokens per question.
End-to-end latency of 150 to 250 milliseconds, with median model compute of 50 to 100 milliseconds.
Question types include yes/no, multiple-choice with 2 to 255 options, and rubric or ordered scoring.
Output includes calibrated confidence scores from 0 to 1 and per-option probabilities.

What it means for people building with AI

For developers building agentic coding pipelines, this gives a cheap, low-latency primitive for the routing, guardrail and classification steps that generative text models handle today at higher cost and latency. Because output tokens are not charged and responses land in 150 to 250 milliseconds, teams can run decision checks at each step of an agent without the cost or delay of a full text completion. Support is already built into agent and gateway frameworks on day one, so existing pipelines can adopt the model without building custom integrations.

Availability and integrations

Pricing is $0.10 per 1 million input tokens, with no charge for output tokens.
A free prototyping tier is available in Google AI Studio.
Day-one integrations include OpenRouter, LangChain, Vercel, LiteLLM, Pydantic AI, Pipecat, LiveKit and OpenCode.
Google says support is coming soon to Gemini in Google Cloud (Vertex AI).

Sources

Introducing the Gemini Decisions API (Google, November 11, 2026, official)
AI Studio changelog - Gemini Decision and Decisions API released (Google AI Studio, November 11, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.