Announcement of the Gemini Decisions API
Google has introduced the Gemini Decisions API, a new endpoint designed for fast, closed-set decisions on multimodal inputs. The API is powered by the Gemini Decision 1 model (identified as `gemini-decision-1`) and is available through the Gemini API in Google AI Studio.
The goal of this tool is to provide purpose-built primitives for bounded choices, such as tool selection, policy checks, model routing, or context grading. Unlike standard generative models that produce text token by token, this system scores candidate options in a single forward pass, returning probabilities and confidence scores.
Architecture and decision types
The model processes text, images, audio, video, and PDFs directly in a single pass, allowing modalities to be combined in a shared input. For example, a call recording can be paired with account metadata, or a scanned PDF with damage photos. The system supports up to 256K tokens of context and allows adjusting the resolution of images and videos to control visual detail and token cost.
The API supports three types of questions in a single request: noul (yes/no), choice (multiple choice), and score (rubric scoring). Every answer includes a confidence score between 0 and 1, enabling developers to route ambiguous cases to a human reviewer or a fallback system.
Question types supported by the API
Comparison of the three evaluation types available in the Gemini Decisions API.
| Type | Description | Output |
|---|---|---|
| noul | Estimates the probability that a yes/no condition holds. | Probability and optional true/false criteria. |
| choice | Selects from 2 to 255 options. | Winning option and normalized probabilities for each candidate. |
| score | Grades inputs according to an ordered rubric (starting at 0). | Continuous weighted score and per-level probabilities. |
Performance, latency, and cost
In benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, Gemini Decision 1 exhibits the lowest calibration error in the evaluated suite. By scoring options without token-by-token decoding, the median model compute latency is between 50 and 100 ms, with an end-to-end latency of 150 to 250 ms.
Performance remains nearly flat as the number of questions per request increases, from 1 to 100. In terms of cost, the API bills $0.10 per million input tokens and $0 for output tokens. Google states that this makes it nearly 5 times cheaper than Gemini 3.5 Flash-Lite with the Interactions API across the shared benchmark suite.
Implementation and code examples
The API uses the same Content block format as the Interactions API. Developers can start building immediately by obtaining an API key in Google AI Studio. The official documentation provides guides for Python, JavaScript, and REST.
The following example shows how to check an image of a counter along with a text note against all three question types in a single request using the Python library.
Code to evaluate an image and text with noul, choice, and score questions.
from google import genai
client = genai.Client(http_options={"api_version": "v1beta"})
result = client.decisions.create(
model="gemini-decision-1",
input=[
{
"type": "image",
"uri": "https://storage.googleapis.com/generativeai-downloads/images/spill.jpg",
"mime_type": "image/jpeg",
},
{
"type": "text",
"text": "Breakroom counter check before the 9 AM client meeting.",
},
],
questions={
"has_spill": {
"type": "noul",
"instructions": "Is there an uncleaned liquid spill on the counter?",
},
"next_action": {
"type": "choice",
"instructions": "Which cleanup action fits the scene?",
"criteria": {
"wipe_with_cloth": "Small counter spill with a cloth or towel already nearby.",
"dispatch_janitorial": "Large floor hazard, broken glass, or active leak.",
"ready_for_meeting": "Counter is clean and ready.",
},
},
"severity": {
"type": "score",
"instructions": "Rate the severity of the counter mess.",
"criteria": [
"Clean and ready.",
"Minor localized spill.",
"Major spill or safety hazard.",
],
},
},
)
Internal use cases at Google
Several Google teams are already using the Decisions API as a low-latency control plane. In Gemini Notebook, the API acts as a real-time router that directs incoming queries to the optimal model tier and thinking level, reducing latency by up to 40% for everyday queries.
In Stitch, the design tool, the API generates playable wireframes of a product flow in seconds and applies edits made by voice or tap almost instantly. In CC (from Google Labs), it is used to select the right emoji reaction from a palette of more than 200 options in a fraction of a second. Additionally, Google Search employs it as a safety layer to detect adversarial behavior in MCP tools.
Availability and ecosystem
The API is available today with a no-cost tier for prototyping in Google AI Studio. The Gemini Decision 1 model is billed at $0.10 per million input tokens and does not charge for output tokens, audio transcription, or OCR. The shared input is billed once per request, regardless of the number of questions evaluated.
Ecosystem integrations are rolling out on platforms such as OpenRouter, LangChain, Vercel, LiteLLM, Pydantic AI, Pipecat, LiveKit, and OpenCode, alongside the google-genai SDKs. Support for Gemini in Google Cloud is expected to arrive soon.
Keep exploring
Corrections and transparency
If you spot incorrect or outdated information, send us a correction with the page and source we should review.
Submit a correction