Gemini Decisions API: fast, calibrated multimodal decisions
Google DeepMind launched the Gemini Decisions API powered by Gemini Decision 1, scoring up to 1,000 yes/no, choice, or rubric questions in a single pass over text, images, audio, video, and PDFs at 150–250 ms end-to-end.
Why it matters
Decisions as single-pass scoring instead of token generation: up to 1,000 bounded questions answered in one request for $0.10 per 1M input tokens with zero output tokens billed, aimed at the plumbing agents run all day (routing, policy checks, grading).
What you could build with it
A startup running a multi-agent support system could route every tool choice, policy check, and reply-quality grade through the Decisions API, replacing a generative call per judgment with one batched scoring pass and cutting both latency and cost on high-volume agent traffic.
Does it hold up?
Too early to judge: live now in Google AI Studio with a no-cost tier, but benchmarks are Google's own, Cloud support is still coming, and no independent production usage has surfaced yet.
Built with Gemini Decisions API
- Google Launches Gemini Decisions API for Multimodal AI Workflowsunite.ai · Independent launch coverage: 262K-token context, three question types, and Google's claim of ~5x lower cost than Gemini 3.5 Flash-Lite with the Interactions API.
- Google launches Gemini Decision 1, a model that scores up to 1,000 questions in one passmlq.ai · Notes the API is already used inside Google to route queries in Gemini Notebook and as a safety layer for MCP tools in Search.
- Gemini Decision 1: Google's AI that answers without writingmadrobot.blog · Walkthrough of the pricing and the benchmark context, including that Jev wins two of the shared tests and comes out cheapest.
- Gemini Decisions API: How Small Teams Can Use It to Sort Leads and TicketsDEV · Practical walkthrough of using the API to triage leads and support tickets, with caveats on the v1beta limits.
Learn more
First spotted on article: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.