Google · 2026-11-11 · major
Gemini Decisions API — fast, calibrated answers to closed-set questions
In short
Google's Gemini Decisions API runs the new gemini-decision-1 model to score yes/no, multiple-choice and rubric questions in one forward pass, with confidence scores, 150-250 ms latency and $0.10 per 1M input tokens.
Google · 2026-11-11 · 3 sources checked
A new Gemini API endpoint that scores your options and returns probabilities instead of writing text.
Key specs
| Price (input) | $0.10 / 1M tokens |
|---|---|
| End to end latency | 150-250 ms |
Quick facts
| Maker | Google DeepMind |
|---|---|
| Model | gemini-decision-1 |
| Input limit | Up to 256K tokens, up to 1,000 questions per request |
| Price | $0.10 / 1M input tokens, $0 output |
| Availability | Gemini API and Google AI Studio (free tier) |
| Question types | Yes/No, multiple choice (2-255 options), ordered rubric |
What is it?
The Gemini Decisions API is a new Gemini API endpoint for bounded choices such as tool selection, policy checks, model routing and context grading. It is powered by a new model, Gemini Decision 1 (gemini-decision-1), and returns a probability and a confidence score for each answer instead of generated text.
How does it work?
You send one input (text, images, audio, video or PDFs) with up to 1,000 independent questions. The model scores the candidate options in a single forward pass. Each question is a yes/no, a choice from 2 to 255 options, or an ordered rubric that returns a weighted score. Google reports 150-250 ms end-to-end latency, nearly flat from 1 to 100 questions.
Why does it matter?
Agents often pay for a full generation just to pick a route or pass a safety check. A confidence score lets code send unclear cases to a fallback model or a human. Google says the model is nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API on its shared benchmark suite, and Google Search already uses it as a safety layer for MCP tools.
Who is it for?
Agent builders and teams running high-volume routing or moderation
Frequently asked questions
- How much does the Gemini Decisions API cost?
- The Gemini Decisions API costs $0.10 per 1M input tokens and $0 for output tokens, according to Google. Input is billed once per request no matter how many questions it carries, and Google lists no separate thinking-token, audio transcription or OCR charges. Google AI Studio offers a no-cost tier for prototyping.
- How many questions can one Decisions API request hold?
- One Gemini Decisions API request takes a single multimodal input of up to 256K tokens and up to 1,000 independent questions. Multiple-choice questions can have 2 to 255 options. Google says latency stays nearly flat from 1 to 100 questions per request.
- How does Gemini Decision 1 compare with Gemini 3.5 Flash-Lite?
- Google says Gemini Decision 1 is nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API across a shared suite covering agent routing, reasoning, ranking and multimodal understanding, and that it has the lowest calibration error. The specific scores are published only as images in Google's post.
- Where can I use the Gemini Decisions API today?
- The Gemini Decisions API is available in the Gemini API and Google AI Studio, which also has a playground. Google lists OpenRouter, LangChain, Vercel AI SDK, LiteLLM, Pydantic AI, Pipecat and LiveKit as integrations, and says support for Gemini in Google Cloud is coming soon.
Try it
client.decisions.create(...) with model gemini-decision-1 (genai client, api_version v1beta)