Overview
- Google launched the Gemini Decisions API, which debuted on Wednesday, November 11, as a v1beta endpoint in Google AI Studio with a no-cost tier for prototyping.
- The gemini-decision-1 model accepts a single shared multimodal input and can evaluate up to 1,000 yes/no, multiple-choice, or ordered-rubric questions in one forward pass and returns per-answer probabilities and a 0–1 confidence score.
- Google reports end-to-end latency of about 150–250 milliseconds and pricing of $0.10 per 1 million input tokens with no charges for output tokens, and the company bills the shared input once per request regardless of the number of questions.
- Google says teams are already using the API inside Gemini Notebook, Stitch, CC and Search for model routing, rapid UI flows, emoji selection and safety checks, and ecosystem integrations with tools like LangChain and OpenRouter are rolling out.
- The Decisions API joins similar products from OpenAI and Microsoft and is meant to cut cost and delay for routine gating tasks inside agent and enterprise stacks, though Google’s performance and cost claims are self-reported and await independent validation.