Gemini Decision 1 API

Access Google Gemini Decision 1—also searched as Gemini Decisions API / gemini-decision-1— through EvoLink's unified API on the Decisions endpoint. Test typed answers, image input, low-latency routing before integrating.

GoogleDecision modelAvailable

from $0.100 / 1M input tokens

API docs

Decisions APIYes/no questionsChoice questionsScore questionsAudio, video, and PDF inputInput-only billing

Production routeLive

Live

Context

262K context

Best For

Routing, policy checks, rubric scoring, tool selection

Input

Text + images + audio + video + PDF

Output

Typed answers: probability · choice · score

Models Pricing Readme API

Choose Gemini Decision 1

One model ID on Google's Decisions endpoint. The card lists the input rate, which is the same for text, images, audio, video, and PDFs; output tokens are always zero and are not billed.

Gemini Decision 1

Google's decision model on the Decisions API for calibrated answers at volume.

Selected

Model ID

gemini-decision-1

Best for

Routing, policy checks, rubric scoring, and tool selection where your code needs a probability, a fixed option, or a score instead of generated text

Input

$0.100 / 1M

6.8 cr / 1M

Output

Not billed

All rates are per 1M tokens, shown in USD and credits, and reflect your account's current pricing. On this endpoint only input tokens are billed, at one rate for every input type; output tokens are always zero.

Gemini Decision 1 pricing

Estimate what one Gemini Decision 1 request costs before you integrate. The calculator uses your account's current rates, with official pricing as a reference.

Request calculator

Enter the token mix for one request.

Input tokens

Estimated request cost

Gemini Decision 1

USD$0.0002

Credits0.01

Input tokens0.0034 cr

Minimum charge: 0.01 credits per request.

Budget guide

Approximate requests using the current token mix.

Add credits

$10

About 65000 requests

For quick testing

$50

About 340000 requests

For regular development

$100

About 680000 requests

For production evaluation

Gemini Decision 1 API

Route requests, check content against many rules at once, and score text, images, audio, video, or PDFs with Gemini Decision 1 on Google's Decisions API. Get a calibrated yes probability, a fixed choice, or a rubric score in one synchronous call through EvoLink's unified gateway, using the same API key and input-only billing. The Decisions API is in beta. Evaluate up to 1,000 independent questions per request, within the shared context limit.

Gemini Decision 1 is served on EvoLink under the model ID gemini-decision-1 through Decisions API, with the same API key and balance you use for every other model. It offers a 262K context window, plus typed answers, image input, low-latency routing.

Gemini Decision 1 specs and capabilities

Numbers come from the EvoLink route configuration; capabilities are what the API exposes today.

Context window: 262K tokens
Input: Text + images + audio + video + PDF
Output: Typed answers: probability · choice · score
Protocols: Decisions API
Model ID: gemini-decision-1

Gemini Decision 1 for routing, checks, and scoring

Use Gemini Decision 1 when application logic needs a yes probability, a fixed choice, or a rubric score with a confidence it can threshold. Keep generated replies, field extraction, and tool calls on a Gemini chat model; both are available through the same EvoLink gateway.

Route Requests to the Right Queue, Model, or Agent

Describe each destination as an option, such as billing, technical support, or a human reviewer. A choice question returns the selected option, a probability for every option, and a confidence, so your code can route automatically and escalate low-confidence cases. A choice is returned even when none of the options fits. Add an explicit other option or a separate yes/no scope check, and send uncertain cases for review.

Check Content Against Many Rules at Once

Ask independent yes/no questions in one request, one per policy rule. Each question runs against the same shared input and does not see the others, and the shared input is billed once no matter how many questions you add.

Score Candidates Against a Rubric

Define ordered levels such as cosmetic, workaround available, and fully blocked. A score question returns a probability-weighted score that can fall between levels, plus the probability of each level, so you can rank or filter results.

Decide on Images, Audio, Video, and PDFs

Check whether a product photo shows damage, whether a call recording mentions cancellation, or whether a PDF matches a document type. Pass the file inline as base64 or as a public HTTPS URL next to any text context.

Two ways to use Gemini Decision 1: EvoLink API or Agent

Use the EvoLink API for product backends and batch jobs, or call Gemini Decision 1 from Codex, Claude, or Gemini to label, route, and score items inside a workflow. Both paths share the same EvoLink API key, balance, model ID, and request history.

Option 1

Integrate with the EvoLink API

Best for: product backends, batch jobs, automated pipelines

Send Gemini Decisions requests to EvoLink in Google's native format: the evidence goes in input, and a questions map keyed by your own IDs declares each question's type, instructions, and criteria.

1Create an EvoLink API key in the console
2Point the google-genai SDK at the EvoLink base URL with api_version v1beta, or POST to /v1beta/decisions, with the model ID shown above and your key in the x-goog-api-key header
3Send one representative request and read usage.total_input_tokens; it is the only billed quantity
4Give every question a short key and set your own thresholds on the returned confidence and probabilities

Create EvoLink API Key

Option 2

Call it with an Agent

Best for: labeling, routing, and scoring steps in Codex, Claude, and Gemini

Give the Agent the evidence, the questions, and the acceptance criteria. It assembles the request, calls Gemini Decision 1 through EvoLink, and returns the typed answers with token usage.

1Set EVOLINK_API_KEY in your local environment; never put it in code or a prompt
2Describe the evidence to evaluate and the questions to ask, with choices or score levels
3Ask the Agent to call Gemini Decision 1 through EvoLink and show the request before sending
4Let the Agent report each answer, the input token usage, and any error body

Gemini Decision 1 API code example and error handling

This example shows the shortest runnable request: one piece of text evidence and one choice question sent to the Decisions endpoint. Open the API tab for the complete parameter and response reference.

View complete API docs

cURL

curl -X POST "https://api.evolink.ai/v1beta/decisions" \
  -H "x-goog-api-key: $EVOLINK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-decision-1",
    "input": "I was charged twice for my order.",
    "questions": {
      "department": {
        "type": "choice",
        "instructions": "Which department should handle this request?",
        "criteria": {
          "billing": "Payments, invoices, and refunds.",
          "technical": "Problems using the product.",
          "other": "Anything outside these categories."
        }
      }
    }
  }'

# answers.department.choice is the selected option, answers.department.probabilities
# lists every option, and confidence is the probability of the top option.
# Only usage.total_input_tokens is billed. The endpoint is synchronous and
# rejects fields it does not define, such as stream and temperature.

Unknown or missing parameter

The endpoint rejects fields it does not define. Remove stream, temperature, and other generation parameters, and check that model, input, and questions are present.

Media URL or file reference rejected

Send media as inline base64 data or a public or signed HTTPS URL up to 15 MB with its real MIME type. Files API URIs from your own Google project, gs:// URIs, and YouTube links are not accepted.

Authentication or balance issue

Send your EvoLink API key in the x-goog-api-key header (or as a Bearer token) and confirm the available balance in the console.

Invalid question key or option count

Question keys must be 1–64 letters, digits, underscores, or hyphens. Choice questions need 2–255 options and score questions 2–255 levels.

Rate limited (429)

Back off and retry with jitter; batch or queue requests instead of sending parallel bursts.

Before you take Gemini Decision 1 to production

Four checks that decide cost and reliability.

Send only the documented fields

Requests carry model, input, and a questions map. The endpoint rejects unknown fields such as stream and temperature with a 400 error.

Required

Pass media as base64 or an HTTPS URL

Use inline base64 data or a public or signed HTTPS URL up to 15 MB, with the real MIME type. Files API URIs from your own Google project, gs:// URIs, and YouTube links are not accepted.

Required

Write the full question in instructions

The model never sees your question keys, so put the whole question and any option descriptions in the text. Choose thresholds from labeled examples and the cost of false positives and false negatives.

Recommended

Budget from usage.total_input_tokens

The shared input counts once per request and each question adds only its own instructions and criteria. A minimum charge applies per request, and repeated input across requests counts in full.

Recommended

Input-only billing

Only input tokens are billed, at the same rate for text, image, audio, video, and PDF tokens. Output tokens are always zero, and each request has a minimum charge.

What you can send

Text, images, audio, video, and PDFs, inline as base64 (up to 100 MB per request, 50 MB for PDFs) or as public HTTPS URLs up to 15 MB each, with up to 262K tokens across input and questions.

Gemini Decision 1: choosing a Decisions endpoint

EvoLink

Compare the two Decisions endpoints on EvoLink, using EvoLink prices per 1M tokens.

Feature	Gemini Decision 1	GPT-6 Luna Decisions
EvoLink price / 1M (input / output)	$0.1 / $0	$0.1 / $0
Context window	262K	1.05M
Input types	Text, images, audio, video, PDF	Text and inline images
Best for	Routing, policy checks, rubric scoring, and tool selection where your code needs a probability, a fixed option, or a score instead of generated text	Teams already on OpenAI's request format, and inputs above 262K tokens

Gemini Decision 1

EvoLink price / 1M (input / output)$0.1 / $0

Context window262K

Input typesText, images, audio, video, PDF

Best forRouting, policy checks, rubric scoring, and tool selection where your code needs a probability, a fixed option, or a score instead of generated text

GPT-6 Luna Decisions

EvoLink price / 1M (input / output)$0.1 / $0

Context window1.05M

Input typesText and inline images

Best forTeams already on OpenAI's request format, and inputs above 262K tokens

Prices reflect your account's current pricing; signed-out visitors see Default rates. Both endpoints bill input tokens only, so their output price is zero.

Gemini Model Family

Compare all Gemini models

Gemini 4 Carbon

Google’s most capable Gemini 4 model, with low, medium, and high thinking levels at the same EvoLink price as Gemini 4 Argon.

View model

Gemini 4 Argon

Google’s Gemini 4 model, with a 1M-token output limit and a lower rate for cached input.

View model

Gemini 3.8 Flash

The newest Flash generation for stronger coding, reasoning, and agent workloads.

View model

Other text models on EvoLink besides Gemini Decision 1

GPT-6 Luna Decisions

OpenAI's Decisions API on GPT-6 Luna: predicate, choice, and score answers over text and inline images, billed on input tokens only.

View model

DeepSeek V4.1 Flash

DeepSeek Flash route with text and image input, a 1M context window, and configurable thinking.

View model

GLM-5.3 Flash

Z.ai's multimodal volume tier of the GLM-5.3 generation, with a 1M-token context window.

View model

Grok 4.5

xAI reasoning and tool-use model with a 500K context window and metered server-side tools.

View model

Gemini Decision 1 API - FAQ

Which model ID and endpoint should I use?

Use gemini-decision-1 with POST /v1beta/decisions and send your EvoLink API key in the x-goog-api-key header. The google-genai SDK (version 3.2 or later) works when its http_options set base_url to EvoLink and api_version to v1beta; then call client.decisions.create. If you change only the base URL, the SDK uses the wrong API version and the request fails. This model uses the Google Decisions request format, not OpenAI Chat Completions or Gemini generateContent.

How much does Gemini Decision 1 cost on EvoLink?

The default rate is $0.100 per 1M input tokens, the same for text, images, audio, video, and PDFs. Output tokens are always zero and are not billed. Each request has a minimum charge. Logged-in users see the rate for their own account.

What question types are supported?

Three: noul returns the probability that a yes/no statement holds; choice picks one of 2 to 255 options and returns a probability for each; score rates the input against 2 to 255 ordered levels and returns a probability-weighted score with a legend. One request can mix types and ask up to 1,000 questions, and each question is answered independently.

What does the confidence value mean?

Confidence is the estimated probability that the top answer is right: the probability of the chosen option or level, or the larger of the yes and no probabilities. Use it to decide whether to act automatically or send the case for review.

Can I send images, audio, video, or PDFs?

Yes. Add image, audio, video, or document blocks to input with inline base64 data or a public or signed HTTPS URL, and set the real MIME type. Files API URIs from your own Google project, gs:// URIs, and YouTube links are not accepted through EvoLink. Media tokens are billed as input.

Does it support streaming, batch, or context caching?

No. The Decisions endpoint is synchronous and returns all answers in one response. Batch, context caching, and Flex or Priority inference are not available, so repeated input across requests is billed in full each time.

When should I use a Gemini chat model instead?

Use Gemini Decision 1 when your code needs a probability, a fixed option, or a score to branch on. Use a Gemini chat model such as Gemini 4 Carbon or Gemini 3.8 Flash when you need written replies, fields extracted into your own JSON schema, tool calls, or multi-step reasoning.

How are invalid or blocked requests handled?

Requests that fail validation return a 400 error whose message names the field or rule, and they are not charged. A reference to a file EvoLink cannot read returns 403, and a URL that cannot be fetched returns 404. Input blocked by safety filters returns an error with a code such as safety instead of answers.

Is the Gemini Decisions API in beta?

Yes. Google currently lists the Decisions API as beta under /v1beta. Evaluate question wording, thresholds, and error handling against your own labeled examples before expanding usage, and keep a review path for low-confidence answers.