Google launches Gemini Decisions API for high-speed multimodal classification
In short: Google has introduced the Gemini Decisions API, powered by the new gemini-decision-1 model. Unlike standard generative models, this tool performs single-pass classification and scoring across text, audio, image, video, and PDF inputs, returning confidence scores instead of text.
This summary was generated automatically by AI from Google AI's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1New endpoint for yes/no, multiple choice, and rubric scoring.
- 2150-250 ms end-to-end latency.
- 3Zero output token billing.
- 4Up to 256K token context window.
| Parameter | Before | Now |
|---|---|---|
| Context window | Not specified | 256K tokens |
| Input price | Not specified | $0.10 per 1M tokens |
| Output price | Not specified | $0 |
| Latency | Not specified | 150-250 ms end-to-end |
Why it matters
This API allows you to offload decision-heavy tasks—such as intent routing, policy enforcement, and tool selection—to a specialized model. It eliminates the need for generative token streaming, resulting in faster performance and lower costs for complex agentic workflows.
What it means for AI agents and contact centers
If you build voice agents, test this for real-time intent classification and routing during live calls. Because it supports audio inputs directly and provides confidence scores, it can replace more expensive generative calls for simple path-finding logic, significantly reducing latency.
🧪 Worth Testing
The ability to route agent logic or detect specific call events in under 250ms with zero output costs makes this highly efficient for complex multi-step voice interactions.
Sources
- Google AIOfficialPrimary sourceOriginal article →„Gemini Decisions API: Fast, calibrated multimodal decisions“11 Nov 2026, 20:00
- Published by source
- 11 Nov 2026, 20:00
- Found by our system
- 11 Nov 2026, 20:36
- Summary generated
- 11 Nov 2026, 21:38
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy