Google DeepMind on November 11 released the Gemini Decisions API, powered by a new model called Gemini Decision 1 that evaluates up to 1,000 closed-ended questions in a single pass and returns probability scores with confidence values in 150 to 250 milliseconds4.
The model processes text, images, audio, video, and PDFs, making it the broadest-modality entrant in a product category that has emerged in recent weeks2. The API supports three question types: yes/no, multiple choice, and rubric-based scoring.
Pricing and competitive field
Gemini Decision 1 is priced at $0.10 per million input tokens with no output-token charge, which Google says is about five times cheaper than Gemini 3.5 Flash-Lite. That matches the price of OpenAI's GPT-6 Luna decision model. Jev, the startup whose product helped define the category, remains the cheapest at 4.2 cents per million input tokens. Cloudflare and Microsoft have also shipped similar offerings.
Modality coverage differs across the three established entrants. Jev handles text and JSON only; OpenAI accepts text and images; Gemini Decision 1 covers text, images, audio, video, and PDFs. Google's model borrows terminology from Jev, reusing the term "Noul" for a yes/no question.
Behavior and internal use
In testing by Simon Willison, Gemini Decision 1 returned calibrated probabilities rather than the hard 1.0 or 0.0 scores that Jev and OpenAI produce for yes/no questions. Willison passed a video file with four questions and received probability values such as 0.04976021841520312 for a "dragon" query and 0.9234055387792485 for a "robot" query.
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. The API is available now through Google AI Studio, with integrations for OpenRouter, LangChain, Vercel, and other platforms rolling out.
ANALYSIS The calibrated-probability behavior distinguishes Gemini Decision 1 from competitors that return binary confidence values, potentially giving developers finer-grained signal for downstream routing or filtering logic. Google's multimodal coverage and its price parity with OpenAI's offering leave Jev's cost advantage as the startup's primary differentiator in a category that now includes four major-platform entrants.