Google Launches Gemini Decisions API
Google has launched the Gemini Decisions API, a service that skips the text and returns a ready answer instead: yes/no, a choice among options, or a score from 0 to 1.
Google has released the Gemini Decisions API, a tool powered by a new model called Gemini Decision 1. Instead of writing out an answer in words, it returns a direct result: yes or no, a choice among options, or a score from 0 to 1. Google says each answer arrives in 150 to 250 milliseconds. It costs 10 cents per million input tokens. The company says that is nearly five times cheaper than running the same task through its existing Gemini 3.5 Flash-Lite model.
For companies building AI agents, software that carries out multi-step tasks on its own, this adds a new kind of building block. Until now, a simple check or choice often meant training a separate classifier or sending the request through a full chat-style model. Google says it already uses this service inside its own products: Notebook, a research tool, Stitch, a design tool, and Search. The release signals that major AI labs are starting to sell not just chat models, but narrow tools built for the small decisions agents make along the way.
Source: blog.google
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles