Checked 12 November 2026, Europe/London. One new development meets today’s practical-use threshold.
Google adds an API for fixed-choice decisions
Google launched the Gemini Decisions API on 11 November, using Gemini Decision 1 through Google AI Studio. It evaluates yes/no checks, predefined choices and scoring rubrics across text, images, audio, video and PDFs, returning probabilities rather than a conventional written answer.
Google lists US$0.10 per million input tokens with no output-token charge, plus a no-cost prototyping tier. Shared input is billed once per request; question instructions and criteria add input tokens. Limits include up to 1,000 independent questions and a 256,000-token input. Google Cloud support is described as coming soon. Google’s speed and calibration assertions are vendor claims, not Tecrave test results.
Tecrave analysis: a small service firm with a developer could assess it for sorting enquiries into “quotation”, “existing job” and “other”. Test against labelled examples, keep an ambiguous-case review queue, and measure wrong routing before connecting it to live actions. A confidence score is not a guarantee of correctness. This requires integration work; it is not a ready-made inbox service, and the announcement does not specify the no-cost tier’s quota.
Prepared with AI assistance from linked primary sources. No hands-on testing or independently measured savings are claimed.