One request can carry up to 1,000 questions. Google says results come back in about 150–250 milliseconds end to end — roughly 0.2 seconds — and output is not billed. On 11 November 2026 it launched the Gemini Decisions API, powered by Gemini Decision 1 (gemini-decision-1), in Google AI Studio.
Unlike a chat model, it scores your options in one pass and returns probabilities plus a 0-to-1 confidence score. Input can be text, images, audio, video, or PDFs, plus up to 1,000 questions, in a 262,144-token context window. Types: yes/no, single choice (2–255 options), and a graded rubric starting at 0.
Paid pricing is $0.10 per million input tokens, with output free. Shared input is billed once per request, with a no-cost tier. Google says this is nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API on its test set. The 150–250 ms figure is for contexts up to 16K tokens, nearly flat from 1 to 100 questions, not 1,000. Start in Google AI Studio (v1beta). OpenRouter and other integrations are officially rolling out; Gemini in Google Cloud is coming soon. We'll test it when it lands on OpenRouter. Figures are Google's; we have not tested them.