Gemini Decision
GAA specialized decision/classification model for Google's new Decisions API (client.decisions.create, REST POST /v1beta/decisions), not a text generator: per the model page it 'evaluates typed questions over text, images, video, audio, and PDFs in a single prefill pass and returns calibrated outcome probabilities with zero output tokens' - yes/no (probability 0.01-0.99 plus confidence), single-select over 2-255 options, and rubric rating over 2-255 ordered levels, up to 1,000 independent questions per request over shared input. STATUS SIGNAL, stated so a consumer can weigh it: Google labels this model neither 'Stable'/'GA' nor 'Preview' anywhere we read. status 'ga' rests on Google's own naming convention on the models page ('Model version name patterns', convention as of September 2025: preview model strings carry a -preview suffix, e.g. gemini-2.5-flash-preview-09-2025, while a plain string 'points to a specific stable model'); gemini-decision-1 has no suffix, the model page's Versions field lists only that string, the November 11, 2026 release note says 'Released' with no preview qualifier, and the deprecations page lists it in its own 'Gemini Decision models' table rather than under a preview heading. Note the SDK examples call api_version v1beta. Pricing (paid tier, per 1M tokens): input /usr/bin/bash.10 for text, image, video, audio and PDF alike; output 'Not applicable (0 output tokens)', so price_output_per_mtok is null; context caching 'Not available'; Batch API not supported; free tier input is free of charge. Google bills shared input tokens once per request regardless of question count, and each question adds tokens for its own instructions, criteria and formatting. Token limits from the model page: input 262,144 (context_window); output token limit published as 0 ('up to 1,000 questions per request') - max_output_tokens is left null, following the convention of the catalog's embedding and rerank rows, because the model generates no tokens. Released 2026-11-11 per the API release notes and the deprecations table ('No shutdown date announced', no replacement); the model page's 'Latest update' reads 'November 2026'. No knowledge cutoff is published, so that stays null. First read 2026-11-12, one day after Google's date.
Gemini Decision by Google is tracked live in our LLM catalog, with a 262K (262.144-token) context window. It is generally available (GA).
Last verified: 13 Nov 2026 · sourced from official provider documentation
◎ You're on the watch list. We'll ping you the moment a model launches, changes price, or gets deprecated.
✕ Couldn't subscribe right now — please try again in a moment.
Free forever · no spam · unsubscribe anytime.
Source: Google official documentation ↗
Gemini Decision — questions & answers
What is the context window of Gemini Decision?
Gemini Decision has a 262.144-token context window (262K).
Is Gemini Decision deprecated?
No — Gemini Decision is generally available (GA) and not currently scheduled for deprecation or retirement in our tracker.
Does Gemini Decision support image or vision input?
Yes — Gemini Decision accepts image input. Listed modalities: text, image, video, audio, pdf.
Can I self-host Gemini Decision?
No — Gemini Decision is proprietary. Its weights are not publicly released, so it is available only through Google's API (or hosted partners), not for self-hosting.
What is the API model string for Gemini Decision?
The API model identifier for Gemini Decision is "gemini-decision-1" when calling Google's API.
Track Gemini Decision price & status changes
New models, price cuts, and deprecations — a short email when something actually changes. No spam, unsubscribe anytime.
The last few changes we caught — this is what lands in your inbox:
- New modelGemini 4 Hydrogen launched (GA) - $1.50/$7.50 per MTok, free tier, 1M context
- New modelGemini Decision launched with the new Decisions API - $0.10 per MTok input, zero output tokens; 262K input
- New modelGemini 4 Carbon launched (GA) - $2/$10 per MTok introductory through Nov 30, then $4/$20; 1M context
◎ You're on the watch list. We'll ping you the moment a model launches, changes price, or gets deprecated.
✕ Couldn't subscribe right now — please try again in a moment.
Free forever · powered by the same data on this page.