Google launches Gemini Decisions API with Decision 1 model: closed-set choices, free output tokens
tokumin · x · 2026-11-12
Sundar Pichai announced the Gemini Decisions API, powered by the new Gemini Decision 1 model, now live in Google AI Studio. Instead of generating text token by token, it's built for closed-set choices, scoring up to 1,000 questions across text, images, audio, video and PDFs in a single pass.
Key details:
- Three decision types: yes/no, multiple choice (up to 255 options), and rubric scoring
- $0.10/1M input tokens with output tokens billed at $0 — roughly 5x cheaper than Flash-Lite
- Leads on several benchmark evals with the lowest calibration error
A follow-up notes the demo video Sundar used ran on staging hardware and a pre-release model; the same flow in production is now about twice as fast.
Related event: Google Launches Gemini Decisions API with Gemini Decision 1 Model(24 posts)→
More from Models
- VulnCheck audit: only 0.57% of Anthropic Glasswing's 37,966 findings ever fixed — WeldPond · 2026-11-12
- On PostTrainBench, AI post-training now beats the human-tuned SmolLM3-3B from the same base — maksym_andr · 2026-11-12
- Claude flags mispronunciations in video audio it claims it can't hear — RileyRalmuto · 2026-11-12
- Heavy Users Say Claude's Safety Rails Block Even Basic Chemistry Work, Forcing Fallback to Opus — rbhar90 · 2026-11-12
- Community teams release wave of abliterated Qwen models for 8–24GB local GPUs — udmrzn · 2026-11-12
- Redditor claims Qwen 4 Max slots between DeepSeek v4.1 and Kimi 3.1 Pro in output quality — power97992 · 2026-11-12