Google 发布 Gemini Decision 1 决策模型，单次可评 1000 个问题

Google joins the decision-model wave with Gemini Decision 1, scoring 1,000 questions in a single pass

文章要点

1Google DeepMind 发布 Gemini Decision 1 模型和 Decisions API，专用于是否、多选、评分类闭合判断
2单次推理最多评 1000 个独立问题，覆盖文本、图像、音频、视频和 PDF，150 到 250 毫秒返回概率与置信度
3定价每百万输入 token 0.1 美元且不收输出费用，官方称比 Gemini 3.5 Flash-Lite 便宜约五倍
4官方称内部已用于 Notebook 动态路由和搜索安全检查，Jev、OpenAI、Cloudflare、微软也已推出同类模型

决策要点

Google 把封闭式判断拆成独立模型品类，用概率打分替代生成式回答，主打低延迟、低成本的规模化判定场景。

影响：审核、评分、模型路由这类高频判断任务可重新核算调用成本，官方口径下输入价格约为 Gemini 3.5 Flash-Lite 的五分之一；概率和置信度输出便于按阈值做兜底，但效果目前只有官方数据。
参考：可考虑盘点业务里是与否、多选、打分类的调用清单，估算改用该 API 的成本和延迟；值得在 Google AI Studio 用真实题库小规模对比现有生成式方案的准确率后再决定接入。
依据：价格对比、评测领先和内部用例均为 Google 官方口径，暂无独立验证；文中提及 Jev 先推出同类模型，OpenAI、Cloudflare、微软随后跟进。

正文

Google Deepmind is introducing the Gemini Decisions API, an interface built for fast, closed-ended decisions like yes/no questions, multiple choice, and scale-based ratings. The underlying model, Gemini Decision 1, processes up to 1,000 independent questions in a single pass across text, images, audio, video, and PDFs. It scores options directly and returns probabilities with confidence values in 150 to 250 milliseconds. Google says it already uses the API internally for dynamic model routing in Gemini Notebook, design prototyping in Stitch, and safety checks in Google Search.

Gemini Decision 1 evaluates a photo of a kitchen counter in a single pass using all three question types: yes/no stain detection (confidence 0.95), selection of the appropriate cleaning action (98 percent), and severity rating on a scale (1.19 out of 2). | Image: Google

The API costs $0.10 per million input tokens with no output-token charge, about five times cheaper than Gemini 3.5 Flash-Lite according to Google. It's available now through Google AI Studio, with integrations for OpenRouter, LangChain, Vercel, and other platforms rolling out.

Gemini Decision 1 leads the benchmark suite in categories like agent routing, reasoning, and multimodal understanding while achieving the lowest calibration error (Top-1 ECE). | Image: Google

Similar models have been popping up in recent weeks. After the startup Jev introduced its version, OpenAI, Cloudflare, and Microsoft quickly followed with similar offerings.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

决策要点

AI 评分67

Google 把封闭式判断拆成独立模型品类，用概率打分替代生成式回答，主打低延迟、低成本的规模化判定场景。

影响：审核、评分、模型路由这类高频判断任务可重新核算调用成本，官方口径下输入价格约为 Gemini 3.5 Flash-Lite 的五分之一；概率和置信度输出便于按阈值做兜底，但效果目前只有官方数据。
参考：可考虑盘点业务里是与否、多选、打分类的调用清单，估算改用该 API 的成本和延迟；值得在 Google AI Studio 用真实题库小规模对比现有生成式方案的准确率后再决定接入。
依据：价格对比、评测领先和内部用例均为 Google 官方口径，暂无独立验证；文中提及 Jev 先推出同类模型，OpenAI、Cloudflare、微软随后跟进。