Gemini Decision and the Decisions API: yes/no, choice and score answers with zero output tokens (November 11, 2026)

What changed

On November 11, 2026 Google released the model gemini-decision-1 and the Decisions API. You send it text, images, audio, video or a PDF plus a set of questions in one request. It sends back probabilities, not generated text, so output token usage is always 0.

There are three question types:

• noul checks whether a statement holds. It returns a probability between 0.01 and 0.99, plus confidence.
• choice picks one option from a fixed list. It returns choice, probabilities and confidence.
• score rates the input against ordered levels. It returns score, probabilities, legend and confidence.

The endpoint is POST https://generativelanguage.googleapis.com/v1beta/decisions, on the Gemini Developer API (v1beta) only. It needs google-genai 3.2 or later, for both Python and JavaScript. One request can carry 1 to 1,000 questions, within a 262,144-token context window.

Pricing: input is free on the Free Tier and costs $0.10 per 1M tokens on the Paid Tier. There is no output price. Shared input is billed once per request, however many questions you ask. Free Tier content can be used to improve Google's products; Paid Tier content is not.

Prerequisites

• A Gemini API key in GEMINI_API_KEY
• Python: pip install -U google-genai (3.2 or later)
• Node: npm install @google/genai@latest (3.2 or later)

Steps

1. Create the client on the v1beta API version: genai.Client(http_options={"api_version": "v1beta"}) in Python, or new GoogleGenAI({ apiVersion: "v1beta" }) in JavaScript.
2. Call client.decisions.create(...) with model="gemini-decision-1", your input and a map. Each question key maps to an object with its and .