Google launches Gemini Decisions API for fast AI choices

Google has introduced the Gemini Decisions API, a tool that lets applications choose between predefined options instead of generating a written response. Min Choi and Felix Yu announce the release on Google’s official blog.

The API uses a new model called Gemini Decision 1. It can assess text, images, audio, video and PDFs, including combinations of those formats. Developers can submit up to 1,000 questions about the same input in one request.

For example, an application could examine a photo and a written note to decide whether a counter needs cleaning, select an appropriate action and rate the severity of the mess. The API returns probabilities and confidence scores, which developers can use to send uncertain cases to another model or a human reviewer.

Speed and pricing

Google says the model scores options in a single pass rather than producing answers word by word. For inputs of up to 16,000 tokens, the company reports a total response time of 150 to 250 milliseconds. Tokens are the units used to measure AI input. The API supports inputs of up to 256,000 tokens, though Google does not give the same latency estimate for the largest requests.

The service costs 10 cents per million input tokens, with no charge for output tokens. The shared input is billed once per request, while each question adds its own instructions and criteria to the input count. Google also offers a free tier for testing in Google AI Studio.

Google says its teams already use the API for tasks including routing queries to suitable models and checking for adversarial content in Search tools.