🔍 Read the full analysis: Explore Fast, Calibrated Multimodal Decisions With The Gemini Decisions API on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Google announced the Gemini Decisions API on Nov. 11, powered by Gemini Decision 1, for scoring up to 1,000 defined questions against one shared text, image, audio, video or PDF input. Google reports 150–250 milliseconds of end-to-end latency and input pricing of $0.10 per million tokens, but the announcement does not provide independent benchmarks or detailed calibration results.
Google announced the Gemini Decisions API on Nov. 11, introducing an endpoint powered by its Gemini Decision 1 model that can score up to 1,000 defined questions against one shared multimodal input, as described in the original analysis. The API is aimed at developers building systems that need to make several bounded choices about the same material, such as routing a request or checking a policy condition, rather than generating a separate text answer for each question.
The endpoint accepts text, images, audio, video and PDFs, with a shared input limit of up to 256,000 tokens, according to the announcement. Developers can submit yes-or-no questions, multiple-choice questions with two to 255 options, or ordered rubric questions. Google says a request can include as many as 1,000 independent questions about the shared input.
Responses include confidence values from 0 to 1. Choice questions also return normalized probabilities for each option; rubric questions return probabilities by level and a weighted score. Google says the model scores candidate options in a single forward pass, instead of producing an answer one token at a time. The endpoint is available through the Gemini API in Google AI Studio, which offers a no-cost tier for prototyping.
Google reports 150–250 milliseconds of end-to-end latency, including 50–100 milliseconds for model computation. It says latency stays nearly flat from one to 100 questions per request and describes performance tests involving multimodal inputs and contexts up to 16,000 tokens. Those are company-reported figures; the announcement does not include an independent evaluation or full benchmark results.
Google lists pricing at $0.10 per million input tokens, with no charge for output tokens. It says shared input is billed once per request regardless of the number of questions, and that adding questions has only a small effect on cost. The company also compares the API with Gemini 3.5 Flash-Lite through its Interactions API, claiming the new approach was nearly five times cheaper across a shared benchmark suite. The announcement does not identify that suite or provide enough detail to reproduce the comparison.
Gemini Decisions API Fast, calibrated multimodal decisions
Google’s API scores up to 1,000 defined questions against shared multimodal input. Its speed, pricing and calibration claims await independent evaluation.
How it works
Gemini Decision 1 scores defined answers in one pass, instead of generating a separate text response for each question.
Provide text, images, audio, video or PDFs once for the batch.
Ask yes/no, multiple-choice or ordered rubric questions.
Get answers with confidence values; choice questions also return probabilities.
Where it could help
Several checks on the same material may need fewer repeated model calls.
Choose a path
Route requests to a model tier or fallback when confidence is low.
Check policy
Score defined conditions across documents, images or recordings.
Grade against a rubric
Return ordered scores for context without generating a separate answer for each check.
What the announcement supports
These figures and deployments are reported by Google; the source gives no independent evaluation.
- Latency stays nearly flat from one to 100 questions.
- Gemini Notebook reports up to 40% lower latency for everyday queries.
- Shared input is billed once per request.
- Benchmark names, datasets and detailed results.
- How confidence calibration holds across tasks and customer data.
- Performance across modalities and production workloads.
Try it in context
Developers can prototype through the Gemini API in Google AI Studio’s no-cost tier.
Batching Could Cut Repeated Model Calls
The API offers a different pattern for workflows where many decisions depend on the same material. A system could, for example, evaluate multiple conditions against a document or audio recording in one request, rather than sending the same input through a model repeatedly. If the reported latency and pricing hold in production, that could make some routing, triage and classification tasks faster or less costly.
Google says Gemini Notebook uses the endpoint to route queries and reports up to 40% lower latency for everyday queries. It also names Stitch, CC from Google Labs and a safety layer for MCP tools in Search as uses. These are company-reported deployments, not independently measured customer results; the announcement supplies no detailed methods or results for those examples.
Confidence values could help a developer send uncertain cases to a fallback system or a human reviewer. But a score is not proof that a decision is correct or safe. The practical value depends on how well those values are calibrated for a specific task and data, and whether the service behaves consistently under real workloads.
Products Worth Considering
Jackery Explorer 300 Portable Power Station with 40W Solar Panel
DJI Mini 3 Drone with DJI RC and 3-Axis Mechanical Gimbal
Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB‑C Charging, Up to 2X More Active Noise Cancelling, Transparency Mode, Adaptive
As an affiliate, we earn on qualifying purchases.
A Scoring Endpoint, Not a Chat Reply
Many generative model APIs produce responses as sequences of output tokens. Google presents Gemini Decision 1 as an option for a narrower problem: assigning scores to a known set of candidate answers in one pass. This structure may suit applications where the available actions or categories are already defined, rather than tasks that call for an open-ended written response.
The launch materials describe three question formats: binary conditions, selection among a fixed set of options, and ordered rubric scores. That makes the endpoint’s role distinct from a general-purpose conversational interface, even though it can evaluate multimodal material. The announcement’s examples include selecting a model tier, checking a policy condition, routing a request and grading context.
Google’s cost comparison with Gemini 3.5 Flash-Lite is presented as evidence for the endpoint’s efficiency, but the available information leaves important comparison details out. The company does not name the benchmark workloads, publish scores or explain the cost assumptions. The claim should be treated as a vendor comparison until its methods and results can be examined.
“Rather than generating answer text token by token, Gemini Decision 1 scores your candidate options in a single forward pass.”
— Google, in its announcement
Benchmark and Calibration Data Await
The announcement does not provide benchmark names, datasets, scores, confidence intervals or an independent assessment of the claimed performance and calibration. It also does not explain how the reported latency varies across each supported modality, or provide detailed measurements for the claimed cost advantage. Without those details, developers cannot independently compare the claims with their existing systems.
Google says answers include confidence values from zero to one, but the launch material does not explain how calibration was measured or whether scores remain reliable across different tasks and customer data. It is also unclear how the service handles ambiguous inputs, how latency changes under heavier traffic, or how performance varies when requests approach the 1,000-question limit. These are relevant questions for teams considering use in safety checks or other consequential workflows.
Production Testing Will Test the Claims
Developers can start testing the endpoint through the Gemini API and Google AI Studio, including the studio’s no-cost prototyping tier. The next useful evidence would include published benchmark methods, calibration evaluations across tasks and modalities, and performance data from production use.
Teams considering deployment can measure response times and costs against their current classifiers or generative model calls, then test whether confidence scores identify cases that need review. Google has not announced a date for additional benchmark disclosures or another product milestone in the source material. For now, the reported speed, pricing and deployment results remain Google’s claims rather than independently verified findings.
Key Questions
What does the Gemini Decisions API do?
It lets developers submit up to 1,000 defined questions about one shared input and receive scores. The questions can be yes-or-no, multiple-choice or based on an ordered rubric.
What kinds of input does the API accept?
Google says the endpoint accepts text, images, audio, video and PDFs. Its announced shared-input limit is up to 256,000 tokens.
How much does it cost?
Google lists a price of $0.10 per million input tokens and no charge for output tokens. It says the shared input is billed once per request, regardless of how many questions are evaluated.
Are the speed and calibration claims independently verified?
Not in the launch material. Google reports 150–250 milliseconds of end-to-end latency, but the announcement provides no independent benchmark results or detailed calibration methodology.
Where can developers try it?
The endpoint is available through the Gemini API in Google AI Studio. Google AI Studio also has a no-cost tier for prototyping.
Primary source: Google AI · via ThorstenMeyerAI.com
