Gemini Decisions lets AI choose between options: what it means for agents
This is not a new chat interface but a component that assigns probabilities to bounded choices: Google announced the Gemini Decisions API on November 11, 2026, with Gemini Decision 1 available through the Gemini API in Google AI Studio.[1]
It is yesterday’s announcement, not a product launched today; it has not been covered in our previous drafts.
## One checklist over shared material
Google says a request can combine text, images, audio, video and PDFs with up to 1,000 independent questions about the same input.[1]
Supported formats include yes/no checks, multiple choice with two to 255 options, and ordered-rubric scoring; answers include confidence scores.[1]
Described uses include tool selection, model routing, policy checks and context grading.[1]
## The useful limitation
VentureBeat explains that questions are evaluated independently: if one decision depends on another, the application must manage that dependency in code or make a follow-up call.[2]
Its analysis distinguishes bounded choices from generating an explanation or extracting an arbitrary invoice number, which require a separate step.[2]
An application can use confidence scores to route ambiguous cases to a person or another model rather than automatically turn every answer into an action.[1][2]
## Pricing and performance are vendor statements
Google quotes $0.10 per million input tokens with no output-token charge and a no-cost tier for prototyping in AI Studio.[1]
Shared input is billed once per request, while each question adds its own instruction and criteria tokens.[1]
Launch latency and benchmark comparisons remain Google’s reported measurements: this draft is not an independent test of the service.[1]
## Sources
[1] https://blog.google/innovation-and-ai/technology/developers-tools/gemini-decisions-api — browser-1
[2] https://venturebeat.com/technology/googles-gemini-decision-1-model-scores-up-to-1-000-questions-in-one-api-call — venturebeat