Google Launches Fast, Calibrated Gemini Decisions API for Multimodal Decisions
Google has introduced the Gemini Decisions API, a new endpoint in the Gemini API for Google AI Studio, together with the Gemini Decision 1 model, also called gemini-decision-1. The offering is designed for fast, closed-set multimodal decisions such as tool selection, policy checks, model routing, and context grading. Instead of generating free-form text token by token, the model scores predefined candidate options in a single forward pass and returns probabilities and confidence scores. A single request can combine one shared multimodal input across text, images, audio, video, and PDFs, with inputs of up to 256K tokens, and evaluate up to 1,000 independent questions. Google said the median model compute time is 50 to 100 ms, with end-to-end latency of 150 to 250 ms, and that latency remains nearly flat from one to 100 questions when the shared context is within 16K tokens. The API is priced at 0.10 U.S. dollars per 1 million input tokens, with zero output tokens billed. Google said it is nearly five times cheaper than Gemini 3.5 Flash-Lite using the Interactions API across a shared benchmark suite. The model supports three decision formats in one request: yes/no checks, multiple-choice selection, and rubric scoring. Each answer includes a confidence score from 0 to 1, enabling developers to route ambiguous cases to a fallback model or human reviewer. The input format follows the same Content block structure as the Interactions API, allowing mixed inputs such as a photo, audio transcript, document, and text instructions to be evaluated together in one call. Google positioned the API for applications that need structured decision outputs rather than generated prose, including visual inspection, support triage, compliance review, and agent orchestration. The shared input is billed once per request, while each question adds only its own instruction and criteria tokens. Google said there are no output-token charges, variable thinking-token bills, or separate costs for audio transcription and optical character recognition. Developers can access the service with an API key in Google AI Studio, where a no-cost prototyping tier is available. Google also listed demo apps, including Doorstep Parcels, Speed Sketch, and Find the Zoo Keeper, and provided guides for Python, JavaScript, and REST. Integrations are rolling out across OpenRouter, LangChain, Vercel, LiteLLM, Pydantic AI, Pipecat, LiveKit, and OpenCode, alongside the google-genai SDKs, with support for Gemini in Google Cloud expected soon.
