Gemini Decisions API: Fast, calibrated multimodal decisions

Google introduced the Gemini Decisions API, powered by the new Gemini Decision 1 model, to enable rapid, probabilistic decision-making across text, images, audio, video, and PDFs. The API evaluates up to 1,000 yes/no, multiple-choice, and rubric-scoring questions in a single pass, returning confidence scores without generating answer text. This approach contrasts with traditional methods that require token-by-token decoding, reducing latency and computational overhead. The service is designed for tasks such as routing queries, grading content, or enforcing policy rules efficiently.

The Decisions API provides purpose-built tools for bounded choices like tool selection, policy checks, model routing, and context grading. Users can pass a single multimodal input—up to 256K tokens—along with up to 1,000 independent questions. The API processes these inputs in 150–250 milliseconds end-to-end, with a cost of $0.10 per 1 million input tokens and no charge for output tokens. This pricing model is nearly five times cheaper than comparable benchmarks using other APIs, according to internal evaluations.

Gemini Decision 1 operates without token-by-token decoding, maintaining a median latency of 50–100 milliseconds for model compute and nearly flat performance regardless of the number of questions per request. The model supports multimodal inputs up to 16K tokens and allows tuning of image and video resolution to balance detail and token cost. Confidence scores accompany each answer, enabling automated routing of ambiguous cases to fallback mechanisms or human review. The API uses the same Content block format as the Interactions API for input consistency.

Teams at Google are already leveraging the Decisions API as a low-latency control plane, simplifying feature development by defining options rather than training custom classifiers or relying on generative calls. The API is available today in Google AI Studio, with pricing at $0.10 per 1 million input tokens and no cost for output tokens. Developers can access an API key, documentation, and demo apps to begin building, with ecosystem integrations planned across multiple platforms and SDKs.