Gemini Decisions API: what it is, price and is it free
Google's Gemini Decisions API answers yes/no, multiple-choice and scoring questions in about 150 to 250 ms. What it does, what it costs and who it is for.
Checked and updated 12 November 2026 at 14:35 UTC

Key points
- Google announced the Gemini Decisions API on November 11, 2026, powered by a new model called Gemini Decision 1.
- It does not write text. It picks from fixed options (yes/no, multiple choice or a score) and returns a confidence number from 0 to 1.
- It costs $0.10 per 1 million input tokens with no output charge, and Google AI Studio has a no-cost tier.
The Gemini Decisions API is a new Google service that makes an AI model choose instead of write. You give it some content and a question with fixed options (yes or no, pick one from a list, or score on a scale) and it sends back the answer with a confidence score, typically in 150 to 250 milliseconds. Google announced it on November 11, 2026.
It is a tool for developers, not a new chatbot. But it will sit behind a lot of apps you use, so it is worth knowing what it does.
What it does
The API runs on a new model, Gemini Decision 1 (gemini-decision-1). According to Google's announcement:
- Input: text, images, audio, video and PDFs together, up to 256K tokens.
- Questions: up to 1,000 separate questions about that input in one request.
- Question types: yes/no, multiple choice with 2 to 255 options, and rubric scoring on an ordered scale that starts at 0.
- Output: the chosen option plus a confidence score between 0 and 1 for every answer. No written text at all.
A normal AI model builds its reply one word at a time. Google says this one scores all the possible answers in a single pass instead, which is why it is fast: 150 to 250 ms end to end for inputs up to 16K tokens, and nearly the same speed whether you ask 1 question or 100.
What it costs
| Price | |
|---|---|
| Input (text, image, video, audio, PDF) | $0.10 per 1 million tokens |
| Output | $0, the model produces no output tokens |
| Free tier | Yes, in Google AI Studio |
| Batch API and context caching | Not supported, per the pricing page |
The shared input is billed once per request, and each question adds its own few tokens. Google says that across its test set this works out nearly 5 times cheaper than doing the same job with Gemini 3.5 Flash-Lite.
Who can use it
Anyone with a Gemini API key, through the Gemini API and Google AI Studio, with docs for Python, JavaScript and REST. Support in Google Cloud is "coming soon". Google also lists integrations rolling out with OpenRouter, LangChain, Vercel, LiteLLM and others.
Neither the announcement nor the pricing page labels it as a preview, and neither lists rate limits for the free tier.
What it is for
Think of all the small judgments an app makes before it answers you:
- Routing. Is this an easy question for a small model or a hard one for a big model? Google says its Gemini Notebook app uses the API for this and cut waiting time on everyday questions by up to 40%.
- Picking a tool. An AI agent has to decide which tool to call next. That is a multiple-choice question.
- Safety checks. Google says Search uses it as a safety layer to catch bad behavior in connected tools.
- Grading. Does this document answer the question? Score it 0 to 3.
- Small fun things. Google's CC experiment uses it to pick an emoji reaction from more than 200 options.
Decisions API vs a normal Gemini model
| Gemini Decisions API | A normal chat model | |
|---|---|---|
| Writes text | No | Yes |
| Answer format | One of your fixed options | Free-form |
| Confidence score | Always, 0 to 1 | Not by default |
| Typical speed | 150 to 250 ms | Varies with answer length |
| Best for | Sorting, checking, routing, scoring | Writing, explaining, coding |
For the writing side, see what is new in Gemini 4 Carbon.
What it means for you
You will not open this in an app. You will notice it, if at all, as assistants that respond faster and hand off to the right model or tool more reliably. The confidence score matters too: an app can be built to ask you, or a person, when the model is unsure instead of guessing.
If you want to try it without paying, the AI Studio free tier is the place to start. Our list of free AI tools has more options that cost nothing.
AI Tools This Week is not affiliated with Google.
Frequently asked questions
Is the Gemini Decisions API free?
There is a no-cost tier in Google AI Studio for prototyping. The paid tier is $0.10 per 1 million input tokens, and there is no charge for output.
Can I chat with Gemini Decision 1?
No. It never writes sentences. You send it content and questions with fixed options, and it returns the chosen option and a confidence score.
How many questions can one request hold?
Up to 1,000 independent questions about the same shared input, which can be up to 256K tokens of text, images, audio, video or PDFs.
Sources
Related
Gemini 4 Neon: price, Gemini app and Copilot availability
Gemini 4 Neon is in the Gemini app for AI Pro and Ultra, GitHub Copilot Pro+ and the API at $3/$15 per 1M tokens. No free tier.
Gemini Robotics 2.5: what Google's robot AI can do
Google DeepMind's Gemini Robotics 2.5 controls robot arms and hands. What's new, the Mercedes-Benz test, who can use it and what it means for you.
Gemini 4 Hydrogen: what's new, price and who can use it
Google released Gemini 4 Hydrogen on November 12, 2026, a fast model for everyday agent work. Where it is available, what it costs and how it compares.