Google DeepMind has launched the Gemini Decisions API, which allows users to send up to 1,000 independent yes/no or multiple-choice questions in a single request. The underlying model, Gemini Decision 1, accepts text, images, audio, video, and PDFs before returning probabilities and confidence values within 150 to 250 milliseconds. Google states it currently employs this tool internally for dynamic model routing in Gemini Notebook, design prototyping in Stitch, and safety checks in Google Search. The service costs $0.10 per million input tokens with no charge for output tokens, making it approximately five times cheaper than Gemini 3.5 Flash-Lite. Access is immediate via Google AI Studio, with integrations for OpenRouter, LangChain, and Vercel following shortly. This release arrives after recent introductions from Jev, OpenAI, Cloudflare, and Microsoft.
The primary significance lies in the ability to process large volumes of binary or categorical queries rapidly without paying for generated text. This pricing structure specifically targets high-volume classification tasks where traditional generative models charge for every token produced. Developers can now batch thousands of distinct evaluations into one call, reducing latency and cost for applications requiring immediate scoring across diverse media types.
* No output-token fees apply
* Supports text, images, audio, video, and PDFs
* Returns confidence values alongside probabilities
