Google Launches Gemini Decisions API to Route Decisions Without Generating Text

Google has released the Gemini Decisions API, powered by its new Gemini Decision 1 model, which assigns probabilities to predefined options instead of generating text token by token. Developers can evaluate up to 1,000 questions in one request, score multimodal inputs against rubrics, and use confidence scores to decide when to proceed automatically or escalate for review. Google reports 150 to 250 millisecond end-to-end latency and prices input at $0.10 per million tokens, with no separate output charges, weeks after a competing OpenAI service.

Why it matters: Developers building AI applications can now handle routine classification and yes/no decisions faster and more cheaply than with generative models, and the confidence scores give a practical way to route uncertain cases to human review.