Independent newsroom

Thursday, November 12, 2026

The File

Every claim traces to a primary source. Uncertainty is labeled. Corrections are public.

Labs · Thursday, November 12, 2026

Google opens an API that scores fixed choices

The File desk · Nov. 12, 2026, 6:41 a.m. ET

Partially confirmed1 primary sourcesAdded 6:41 a.m. ET, Nov. 12

Status

Partially confirmed. Google's November 11, 2026 post introduces the Gemini Decisions API and the Gemini Decision 1 model. The latency, the price comparison, and the benchmark rank are the company's. This desk did not run the model.
  • Confirmed
    The post's visible line is Nov 11, 2026. It names the Gemini Decisions API, the model Gemini Decision 1, the model id gemini-decision-1, an input of up to 256K tokens, up to 1,000 questions, yes/no, choice, and rubric questions, and a confidence score.
  • Partially confirmed
    The 150-250 millisecond latency, the $0.10 price, the claim of no output charge, and the comparison with another Google model are Google's. This desk did not reproduce them.
Google on Wednesday, November 11, 2026, introduced a tool that scores a fixed list of answers instead of writing a reply one piece at a time.1 The company calls it the Gemini Decisions API, and the model behind it is Gemini Decision 1.1
"Today, we’re introducing the Gemini Decisions API, a new endpoint for fast, closed-set decisions powered by our new Gemini Decision 1 model (gemini-decision-1), available through the Gemini API in Google AI Studio," the post says.1 In a closed-set decision, the developer lists the allowed answers, and the model scores those answers in one pass. A chatbot instead writes new text token by token, and a token is a small piece of the input the company counts.
The input can run to 256K tokens, across text, images, audio, video, and PDFs.1 A developer can attach up to 1,000 separate questions to that same input.1 A question can be a yes or no, a choice from a list, or a score on a rubric the developer writes.1
Google says the price is $0.10 per million input tokens, and that it does not bill for the output.1
Why it matters (Estimate). Estimate. A developer who needs an app to sort yes-or-no questions, pick from a menu, or grade against a rubric can call this instead of paying a chatbot to write a paragraph. Google says tools it already runs, including Gemini Notebook, use the API to route work. The speed, and the claim that it costs less than another Google model, are Google's account. The post says support in Gemini for Google Cloud is coming soon.
Each answer includes a confidence score, a number that tells the developer's code how sure the model is, so an unsure case can be sent to a person.1
How we know

Notes

  1. 1.Google, "Gemini Decisions API: Fast, calibrated multimodal decisions," visible line Nov 11, 2026. https://blog.google/innovation-and-ai/technology/developers-tools/gemini-decisions-api/ The page's structured data lists datePublished 2026-11-11T18:00:00+00:00, which is 1:00 p.m. ET. The visible byline does not print an hour. Byline: Min Choi, Product Manager, Google DeepMind, and Felix Yu, Principal Scientist, Google DeepMind. "Today, we’re introducing the Gemini Decisions API, a new endpoint for fast, closed-set decisions powered by our new Gemini Decision 1 model (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 (up to 256K tokens) along with up to 1,000 independent questions. Rather than generating answer text token by token, Gemini Decision 1 scores your candidate options in a single forward pass, returning probabilities and confidence scores in 150-250 ms end-to-end latency at $0.10 per 1M input tokens with zero output tokens billed." "Every answer includes a confidence score (from 0 to 1) so your code can route ambiguous cases to a fallback or human reviewer." The post names three question types: yes/no, choice ("Selects from 2 to 255 choices"), and an ordered rubric score. "With zero output tokens billed, Gemini Decision 1 with the Decisions API is nearly 5x cheaper across the shared benchmark suite than Gemini 3.5 Flash-Lite with the Interactions API." A separate latency line, not used above: median model compute of 50-100 ms, and 150-250 ms end to end, "across multimodal inputs and contexts up to 16K tokens." Google says Gemini Notebook uses the API "to serve everyday queries with up to 40% lower latency." It also names Stitch, CC from Google Labs, and a safety layer for Search. "Gemini Decision 1 is priced at $0.10 per 1M input tokens and $0 for output tokens, with an API no-cost tier you can use in Google AI Studio for prototyping." "Ecosystem integrations are rolling out" across named tools, "with support coming soon to Gemini in Google Cloud." Those comparisons and product claims are Google's. Retrieved 6:25 a.m. ET, Nov. 12.

What is still unknown or disputed

  • A clock on the visible byline. The page prints Nov 11, 2026, and does not print an hour there.
  • How the model scores outside Google's tests. This desk did not run it.
  • When Gemini in Google Cloud will offer it. The post says support is coming soon.

Primary sources

Every claim in this story is drawn from the documents below. If a fetch failed, that is recorded on the card.

  1. Source 1

    Gemini Decisions API: Fast, calibrated multimodal decisions

    Google · November 11, 2026

    Google said on November 11, 2026, that the Gemini Decisions API, powered by Gemini Decision 1, scores fixed options in one pass. Input up to 256K tokens, up to 1,000 questions, $0.10 per million input tokens, and no output tokens billed, on the company's account.

    https://blog.google/innovation-and-ai/technology/developers-tools/gemini-decisions-api/