Google Launches Gemini Decisions API
Original: Gemini Decisions API
Why This Matters
Batch multimodal evaluation at 1,000 questions per pass could cut AI pipeline costs significantly.
Google DeepMind announced the Gemini Decisions API on Nov 11, 2026, powered by the new Gemini Decision 1 model. It evaluates up to 1,000 yes/no, choice, and rubric-scoring questions in a single pass across text, images, audio, video, and PDFs.
Google DeepMind has unveiled the Gemini Decisions API, a new developer tool built on the Gemini Decision 1 model. The API is designed to handle batch evaluation at scale — processing up to 1,000 questions per pass — across five modalities: text, images, audio, video, and PDFs.
The offering targets a specific use case: fast, calibrated multimodal decision-making. Question types supported include yes/no, multiple-choice, and rubric-based scoring, suggesting the API is aimed at automated evaluation pipelines, content moderation, quality assurance, and similar classification workflows where speed and consistency matter more than open-ended generation.
The announcement comes from Min Choi, Product Manager, and Felix Yu, Principal Scientist, both at Google DeepMind. The article content is sparse on pricing and latency benchmarks, but the emphasis on "single pass" processing across rich media types signals a focus on throughput efficiency. No third-party comparison data was provided.