Use Gemini’s Decisions API for fast bounded choices
Google launched the Gemini Decisions API for fast closed-set evaluation across text, images, audio, video and PDFs. One request can score up to 1,000 yes/no, multiple-choice or rubric questions and return confidence values without generating answer text token by token.
Source checked Nov 11, 2026
EDITORIAL TAKEAWAY
Use this for routing, grading or policy checks where the allowed outcomes are known in advance. Set a confidence threshold that sends ambiguous cases to a stronger model or human reviewer instead of treating every top score as certain.
Before you try it
- Gemini Decision 1 is available through the Gemini API in Google AI Studio.
- A request can share one multimodal input across text, images, audio, video and PDFs, with up to 1,000 independent questions.
- Supported question types are yes/no probability, multiple choice with 2 to 255 options, and ordered rubric scoring.
- Google lists pricing at $0.10 per million input tokens and $0 for output tokens, plus an API no-cost tier for prototyping.
- The announcement reports 150–250 ms end-to-end latency; actual latency and calibration will vary by input and workload.
- Google Cloud support is coming later, while several ecosystem integrations are still rolling out.
A small next step
Our suggested way to explore the change.
- Choose one bounded decision with known acceptable outcomes, such as routing a request to a model tier.
- Define explicit criteria for every choice and include a fallback for low-confidence results.
- Test the endpoint with representative text or media in the Google AI Studio no-cost tier.
- Compare confidence scores with reviewed answers before placing the decision in an automated workflow.
Go to the source
Google · Gemini Decisions API: Fast, calibrated multimodal decisions ↗Google AI Studio · Decisions API documentation ↗Availability can change. Check the maker's current requirements before making a decision.