Overview
Gemini Decision 1 is the model behind the Gemini Decisions API, which Google announced on 11 November 2026. It is served in the Gemini API as `gemini-decision-1`. Instead of writing text, it scores a fixed set of options and returns a probability and a confidence score from 0 to 1 for each answer.
One request carries a single multimodal input (text, images, audio, video or PDFs, up to 256K tokens) and up to 1,000 independent questions. Questions can be yes/no, a choice from 2 to 255 options, or an ordered rubric that returns a weighted score with per-level probabilities. Google aims it at bounded choices such as tool selection, policy checks, model routing and context grading.
Google reports 150-250 ms end-to-end latency, with median model compute of 50-100 ms, and says latency stays nearly flat from 1 to 100 questions per request. Google also says the model leads its benchmark suite (agent routing, reasoning, ranking and multimodal understanding) with the lowest calibration error, and is nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API on the shared suite. The scores appear only as images in Google's post.
Pricing is $0.10 per 1M input tokens and $0 for output tokens; input is billed once per request regardless of question count. Google AI Studio has a no-cost tier. Google names Gemini Notebook (model routing) and Google Search (a safety layer for MCP tools) among early users.
| Released | 2026-11-11 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Parameters | Not disclosed |
| Modalities | Text, Image, Audio, Video |
| Status | Generally available |
Pricing
| Input | $0.10 / 1M tokens |
|---|---|
| Output | $0 / 1M tokens |
Input is billed once per request regardless of question count; a no-cost tier is available in Google AI Studio.
Strengths
- Returns probabilities and a 0-1 confidence score per answer, so code can route unclear cases to a fallback or a human
- Up to 1,000 independent questions over one input of up to 256K tokens in a single request
- 150-250 ms end-to-end latency, nearly flat from 1 to 100 questions
- $0.10 per 1M input tokens with no output-token charge
Best for
- Reach for it to route a query to the right model or tool before spending a full generation
- Reach for it for policy and safety checks on text, images, audio, video or PDFs
- Reach for it to grade context or rank candidates against an ordered rubric
How to access
| Provider | Model ID |
|---|---|
| Gemini API (Google AI Studio) ↗ | gemini-decision-1 |
| OpenRouter ↗ | google/gemini-decision-1 |
FAQ
What does Gemini Decision 1 cost?
Gemini Decision 1 costs $0.10 per 1M input tokens and $0 for output tokens in the Gemini API. Input is billed once per request regardless of how many questions it carries, and Google AI Studio offers a no-cost tier for prototyping.
What question types does Gemini Decision 1 support?
Gemini Decision 1 supports three question types: yes/no questions that estimate the probability a condition holds, multiple choice with 2 to 255 options and normalized probabilities, and ordered rubrics that return a weighted score with per-level probabilities.
How fast is Gemini Decision 1?
Google reports 150-250 ms end-to-end latency for Gemini Decision 1, with median model compute of 50-100 ms, and says latency stays nearly flat from 1 to 100 questions per request.