Google enters the race with Gemini Decision 1 for fast decisions across thousands of questions

Google DeepMind has introduced the Gemini Decisions API, an interface for closed-ended questions with yes-or-no answers, multiple-choice options, or ratings on a scale. The Gemini Decision 1 model can process up to 1,000 independent questions in a single pass across text, images, audio, video, and PDFs. It evaluates the options directly and returns probabilities with confidence scores in 150–250 milliseconds. Google is already using the API for dynamic model selection in Gemini Notebook, design prototyping in Stitch, and safety checks in Google Search.

How the API works

Unlike a standard model that generates a detailed response, Gemini Decision 1 evaluates the options directly. For each option, it returns a probability and a confidence score. This approach is designed for tasks with a limited set of answers, such as choosing one of several options or giving a rating on a scale.

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

Source: the-decoder.com

Google says the API costs $0.10 per million input tokens, with no charge for output tokens. The company claims this is about five times cheaper than Gemini 3.5 Flash-Lite.

The API is already available through Google AI Studio. Integrations with OpenRouter, LangChain, Vercel, and other platforms will follow.

Google joins the market

In recent weeks, other companies have also released similar models. Following startup Jev, OpenAI, Cloudflare, and Microsoft introduced their own solutions.

Dataist's take: Gemini Decision 1 stands out for processing up to 1,000 questions in a single pass and returning answers in 150–250 milliseconds. Access through other platforms is still expanding: Google says integrations with OpenRouter, LangChain, and Vercel are still rolling out.