Google introduced the Gemini Decisions API on November 11, 2026. The new endpoint, powered by the Gemini Decision 1 model (gemini-decision-1), is available through the Gemini API in Google AI Studio — and OpenRouter now lists the model as available as well.
Unlike a general-purpose chat model, Gemini Decision 1 is designed for closed-set decisions: yes/no questions, multiple choice and rubric-based scoring. Developers can submit one multimodal input covering text, images, audio, video and PDFs, with up to 256,000 tokens, together with as many as 1,000 independent questions. The model scores the candidate options in a single pass and returns probabilities and confidence scores rather than generating a prose answer. Google suggests using it for tool selection, policy checks, model routing and context grading.
Google prices the model at $0.10 per million input tokens, with no output-token charge. The shared input is billed once per request regardless of the number of questions, although each question's own instructions and criteria add tokens. A no-cost tier is available in AI Studio for prototyping. Google also quotes 150–250 ms end-to-end latency and says the model was nearly five times cheaper than Gemini 3.5 Flash-Lite across its shared benchmark suite; those performance and comparison figures are Google's own.
OpenRouter's model page confirms the same $0.10 / $0 per million input/output tokens and a 262K context window. Requests are forwarded to a single provider without routing. There is an important integration difference: the model runs through OpenRouter's Decisions API, not its OpenAI-compatible chat-completions endpoint, so chat SDK request formats will not work. OpenRouter also states that the model is not currently available through EU in-region routing.