Gemini Decisions API: Fast, calibrated multimodal decisions
What happened
Google introduced the Gemini Decisions API, powered by the new Gemini Decision 1 model, allowing developers to evaluate up to 1,000 yes/no, multiple‑choice, or rubric‑scoring questions in a single pass over multimodal inputs such as text, images, audio, video, and PDFs. The service returns probabilities and confidence scores in 150‑250 ms latency and charges $0.10 per 1 M input tokens with no output‑token cost, making it about five times cheaper than Gemini 3.5 Flash‑Lite on benchmark tests. It supports combined modalities, configurable image/video resolution, and provides three primitive question types (noul, choice, score) with detailed confidence outputs.
Why it matters
For ordinary readers, this means faster, cheaper automated decisions for things like content moderation, policy checks, or routing tasks, potentially reducing reliance on human review. The confidence scores let applications handle ambiguous cases more safely.
Reality check
An automated, sceptical read of the sourceIncremental. The API extends existing Gemini capabilities to batch multimodal decision‑making with cost and latency improvements, but it builds on familiar model‑as‑a‑service patterns rather than a fundamentally new breakthrough.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from Google
Google Search gets a major upgrade with Gemini 4 HydrogenGoogle · 12 Nov 2026Introducing Gemini 4 HydrogenGoogle · 12 Nov 2026Gemini 4 HydrogenGoogle · 12 Nov 2026Try Tabletop: Customizable, collaborative spaces for groups to plan, play and do more together.Google · 11 Nov 2026Experience the art and legacy of Dr. Esther MahlanguGoogle · 11 Nov 2026This page was collected from blog.google and summarised automatically by AIPulsee's pipeline. See how it runs.