Google has released the Gemini Decisions API, an endpoint for choice, routing and classification tasks where the answer is picked from a fixed set: which tool an agent should call, whether a request passes a policy check, which model should handle it, or how usable the retrieved context is. The underlying Gemini Decision 1 model scores every option in a single pass and returns probabilities across three question shapes: yes or no with the probability that the condition holds, a choice of 2 to 255 options with a probability for each, and an open-ended score on a continuous scale. One request accepts up to 256,000 tokens of text, images, audio, video or PDF plus up to 1,000 independent questions, and the content is billed once however many questions are attached.

Google says the model leads on its own agent-routing, reasoning, ranking and multimodal test sets with the lowest calibration error, at almost five times less than Gemini 3.5 Flash-Lite through the Interactions API. Pricing is $0.10 per million input tokens with output tokens free.

It is available now through the Gemini API and Google AI Studio, with a free tier for prototyping and Google Cloud support promised later.

Related: OpenAI Opens the Decisions API for Agent Micro-Decisions, Microsoft Releases Decision-1, a Model That Only Makes Decisions, OpenRouter Picks TypeSafe’s Jev to Route Model Selection

Gemini Decisions API