# Google's New Gemini Model Doesn't Write Answers, It Picks Them: Decision 1 Scores Up to 1,000 Questions for $0.10 per Million Tokens

Times of AI Desk · 2026-11-12 · Models

[https://timesof.ai/2026/11/google-gemini-decisions-api-decision-1-closed-set-choices](https://timesof.ai/2026/11/google-gemini-decisions-api-decision-1-closed-set-choices)

> Google launched the Gemini Decisions API on November 11, powered by a new model, Gemini Decision 1, that scores closed-set questions (yes/no, 2 to 255 choices, or rubrics) against one text, image, audio, video or PDF input and returns probabilities and confidence scores instead of generated text. It handles up to 1,000 questions per request at $0.10 per million input tokens with no output charge. Google's own benchmarks put it ahead on five of seven shared tests; the independent Artificial Analysis Decision Index ranks it ninth of 18, per VentureBeat.

Much of what agents do between the big steps is small choices: which tool to call, which model should take a request, whether something breaks a policy. Google now sells a model built only for those. **Gemini Decision 1** does not generate text; it scores a fixed set of options and returns probabilities, which makes it a routing and checking component rather than a chatbot. It also lands in a market that already has rivals, and an independent index does not put it on top.

## What Google launched

Google DeepMind introduced the **Gemini Decisions API** on **November 11**, available through the Gemini API in Google AI Studio with a no-cost tier for prototyping. Per Google:

- **Three question types:** yes/no, multiple choice from **2 to 255** options, and ordered rubric scores. Each answer comes with a confidence score from 0 to 1, so applications can send uncertain cases to a person or another model.
- **One input, many questions:** a single input of text, images, audio, video and PDFs, up to **256K tokens**, with up to **1,000** independent questions per request, scored in one pass.
- **Price and speed:** **$0.10 per million input tokens** and nothing for output. Google reports **150 to 250 ms** end to end; VentureBeat notes that figure covers inputs up to 16K tokens.
- **Availability:** integrations are rolling out across OpenRouter, LangChain, Vercel and others; support in Gemini on Google Cloud is "coming soon."

It extends Google's push to give agent builders supported building blocks, after it made [the Gemini API's computer use tool generally available](https://timesof.ai/2026/11/gemini-api-computer-use-generally-available).

## Google's numbers and an independent check

In Google's launch benchmarks, Decision 1 led five of the seven tests that all four compared models ran, VentureBeat reports, scoring **92.2%** on JevBench against 90.9% for OpenAI's GPT-6 Luna and 90.0% for TypeSafe's Jev 1.14. Jev led on two, including a tool-use test, 82.0% to 79.9%. Google also reported the lowest calibration error of the group.

The independent **Artificial Analysis Decision Index v2.1** ranks Gemini Decision 1 **ninth of 18** at launch, according to VentureBeat, behind models such as Nace AI's Drex Max v1.1 and Jev 1.14. VentureBeat lists OpenAI's own Decisions API, generally available since October 22 and running on GPT-6 Luna, at the same $0.10 base input price. VentureBeat's assessment is that Gemini's case is strongest where one item arrives as a recording, video or document, and that cheaper hosted options exist for short text.

## Limits

- The benchmark wins, latency and calibration figures are Google's. The Artificial Analysis ranking is as reported by VentureBeat; we did not check the index directly.
- Google's and rivals' results come from different runs and settings, so they are not directly comparable.
- Each question is scored independently; tasks that need written explanations or free-form extraction still need a separate model call.

## Sources

- [Google: Gemini Decisions API: Fast, calibrated multimodal decisions (November 11, 2026)](https://blog.google/innovation-and-ai/technology/developers-tools/gemini-decisions-api/)
- [VentureBeat: Google's Gemini Decision 1 model scores up to 1,000 questions in one API call (November 11, 2026)](https://venturebeat.com/technology/googles-gemini-decision-1-model-scores-up-to-1-000-questions-in-one-api-call)