Google releases Gemini Decision model and Decisions API for binary and categorical calls
Google's Gemini Decision model and Decisions API return binary, categorical and rubric decisions for text, images, audio, video and PDFs, and Google says they use zero output tokens.
Google has released Gemini Decision, identified as gemini-decision-1, along with a Decisions API that Google says is built for fast, calibrated decisions. According to Google's November 11, 2026 changelog entry, the API returns binary, categorical and rubric decisions. It accepts text, images, audio, video and PDFs as input, and Google says these decisions use zero output tokens. A binary decision is a yes-or-no answer. A categorical decision picks one label from a set. A rubric decision scores an input against a set of criteria.
The announcement doesn't say how the model reaches its decisions, what it was tested against, or how its calibration compares with other approaches. The changelog includes no prices or rate limits. It links to separate pages for those, along with a Decisions API guide for getting started.
What changed
Google released the Gemini Decision model, identified as gemini-decision-1, and a Decisions API that returns binary, categorical and rubric decisions for text, images, audio, video and PDFs, with zero output tokens according to Google.
Type: New capability · Applies to: gemini-decision-1, Decisions API
Why it matters
If the zero-output-token claim holds for your workloads, teams that now generate text just to get a label or score could cut cost and latency. Buyers should check the published pricing and limits before relying on it.
What you can do
Review the Decisions API guide and pricing page to see whether gemini-decision-1 can replace generation-based classification or rubric scoring in your pipelines.
How the story developed
The changelog gives the model ID gemini-decision-1 and links to the pricing and rate limit pages.
The changelog adds the API model identifier gemini-decision-1 and links to pricing and rate limits.
Adds Artificial Analysis's independent Decision Index score of 66, with cost, latency and benchmark comparisons against Jev and Drex Max.
Adds a hands-on video test where yes/no answers came back as graded probabilities such as 0.92 rather than 1.0 or 0.0, plus a price comparison with GPT-6 Luna and Jev.
Google's own announcement post now confirms the Gemini Decision model and Decisions API, and serves as the primary source.
Original source
- blog.googleGemini Decisions API: Fast, calibrated multimodal decisions
- simonwillison.netGemini Decisions API
- link.mail.beehiiv.comGemini Decision 1
- aistudio.google.comGemini API changelog: November 11, 2026
- aistudio.google.comGemini Decision 1
- aistudio.google.comDecisions API
- aistudio.google.compricing
- aistudio.google.comrate limits
Further reporting
Where we saw it
Gemini API changelog, Google AI, Latent Space, Simon Willison
AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works