What Google announced
The Google blog post, dated November 11, 2026 and written by Min Choi, Product Manager, and Felix Yu, Principal Scientist, both at Google DeepMind, introduces the Gemini Decisions API. It is a new endpoint for fast, closed-set decisions powered by Gemini Decision 1 (gemini-decision-1), available through the Gemini API in Google AI Studio.
You can pass one multimodal input across text, images, audio, video, and PDFs, along with up to 1,000 independent questions. The blog says the input can be up to 256K tokens; the model documentation lists an input token limit of 262,144 and an output token limit of 0. Gemini Decision 1 scores candidate options in a single forward pass and returns probabilities and confidence scores. Google quotes 150-250 ms end-to-end latency at $0.10 per 1 million input tokens, with zero output tokens billed. Those latency and cost comparisons are Google's claims.
Who can use it and when
Google says you can start today with an API key in Google AI Studio and with Python, JavaScript, and REST guides in the Decisions API docs. The official pricing table lists a Free Tier that is free of charge for input, and a Paid Tier at $0.10 per 1 million input tokens, with output not applicable. The blog also mentions a no-cost tier in Google AI Studio for prototyping. Free-tier rate limits in the docs, checked November 12, 2026, are 200 requests per minute, 50,000 tokens per minute, and 10,000 requests per day.
Support in Gemini on Google Cloud is described as coming soon. The post and the docs do not state whether the endpoint is available in the EU or in Poland. Batch API, caching, function calling, and text generation are listed as not supported for this model.
Why it matters
The API is for a bounded choice your code can branch on, such as a policy check, a route between models, or a score against a rubric. Official docs name three question types: yes/no (API type noul), choice (2 to 255 options), and score (an ordered rubric). The docs treat noul as the official type name for a yes/no question. Each answer includes a confidence score from 0 to 1 so code can send unclear cases to a fallback or a person.
Google says Gemini Notebook already uses it as a real-time router, with "up to 40% lower latency" on everyday queries, and mentions uses in Stitch, CC from Google Labs, and Google Search. The "nearly 5x cheaper" comparison with Gemini 3.5 Flash-Lite is also Google's. Chart numbers from the blog graphics are not restated here, because they are not given as figures in the article text.
Access and limitations
Shared input is billed once per request; each question adds only its own instruction and criteria tokens. There are no output-token charges. SDK examples use API version v1beta. AppFor.ai has not called the Decisions API. If you need generated text, thinking, or batch jobs, this is the wrong endpoint. Confirm current quotas and regional access on Google's pages before you ship.
Go to the source
Sources & further reading
Based on the official documentation below. Features and access can change; check the provider before making a decision.
Was this page useful?
Help us make our coverage more useful. This is feedback on the page, not a rating of the tool.
Go a little further