Google launches Gemini Decision 1 calibrated decision model
Original title: [AINews] D-Day (Decision Models Day)
The API scores up to 1,000 closed-set questions per request at $0.10 per million input tokens, with about 200 ms latency.
Latent Space’s AINews recap says Google shipped Gemini Decision 1 and a Decisions API ahead of Anthropic, from an internal DeepMind prototype. The model accepts text, image, audio, video and PDF and returns confidence scores for yes/no, multiple-choice and rubric tasks. Independent Artificial Analysis results tie it with Elephant 9B and Jev 1.14 and trail Drex Max. The same roundup covers Claude Fable 5.2 agent rankings and mixed reports on Gemini 4 Carbon latency.
Unconfirmed: Vendor benchmarks conflict with independent scores, and Qwen-4 quality claims are unconfirmed.
- Output tokens are billed at $0, which Google calls about 5x cheaper than Flash-Lite.
- Google’s tests beat Jev 1.14 on 5 of 7 benchmarks, including 92.2% versus 90.0%.
- Artificial Analysis scores Decision 1 at 66, seven points behind Drex Max v1.1.
- Fable 5.2 ranks first on Agent Arena, with +30.4% across 4.7K sessions.
- Claims that newly live Qwen-4-Max is disappointing remain unconfirmed.
- Vendors
- Products
- Gemini Decisions API
- Organizations
- Artificial AnalysisAnthropic
From the original article
14 min read at latent.space
November 11, 2026 — Google enters the decision model market. At AI Engineer, Cloudflare showed which parts of a build failure a classifier could handle. TypeSafe AI showed off their production customers.
By swyx .ai · latent.space
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.