Google's Gemini API: Smarter, Faster Judgments

Google DeepMind has introduced the Gemini Decisions API, designed for applications requiring quick, sequential judgments. This API can process various data formats, including scanned documents, images, audio, and video, within a single request. It allows software to make up to 1,000 decisions about the same input material in one pass.

The Gemini Decisions API supports different question types: yes-or-no probabilities, multiple-choice selections, and rubric-based scoring. Developers can fine-tune media resolution to manage token consumption. The API's pricing is competitive, starting at $0.10 per million input tokens, with free output. This positions Google against rivals like TypeSafe's Jev and OpenAI's Decisions API.

Google's benchmarks indicate Gemini Decision 1 leads in several decision-making tests, showing strong performance in knowledge and reasoning tasks. The API offers a potential for reduced latency and cost savings, especially for enterprise applications handling complex, multi-format data, aiming to streamline processes like claims intake and content analysis.