Google introduced the Gemini Decisions API on November 11, 2026, as a new endpoint for closed-set decisions, available through the Gemini API in Google AI Studio. A post on The Keyword by Google DeepMind product manager Min Choi and principal scientist Felix Yu says the API is "powered by our new Gemini Decision 1 model (gemini-decision-1)". Developers can send one multimodal input and up to 1,000 questions in a single request. Google prices Gemini Decision 1 at $0.10 per 1M input tokens and $0 for output tokens.

#What changed?

The Decisions API is built for bounded choices such as tool selection, policy checks, model routing and context grading. Google says: "You can pass one multimodal input across text, images, audio, video, and PDFs (up to 256K tokens) along with up to 1,000 independent questions." VentureBeat reported a 262,144-token limit that "covers the shared evidence plus the instructions and choices for all questions."

Rather than generating answer text token by token, Google says, Gemini Decision 1 scores candidate options in a single forward pass and returns probabilities and confidence scores. Google puts end-to-end latency at 150-250 ms. Developers can combine modalities in one shared input, "such as pairing a call recording with account metadata or a scanned PDF claim with damage photos", and can tune image and video resolution to control visual detail and token cost. "Every answer includes a confidence score (from 0 to 1) so your code can route ambiguous cases to a fallback or human reviewer".

The API supports three question types:

  • Yes/no. "Estimates the probability that a yes/no condition holds, with optional true and false description criteria."
  • Multiple choice. "Selects from 2 to 255 choices, returning the winner and normalized probabilities for every candidate."
  • Rubric. "Grades inputs across an ordered rubric (starting at 0), returning a continuous weighted score and per-level probabilities."

VentureBeat reported that "Each question is evaluated independently. If a second decision depends on the first, the application can handle that dependency in code or make a follow-up call." It also noted: "Tasks such as drafting an explanation or pulling an arbitrary invoice number from a file still call for a separate generation or extraction step."

#How much does it cost?

Google prices Gemini Decision 1 at $0.10 per 1M input tokens and $0 for output tokens, "with an API no-cost tier you can use in Google AI Studio for prototyping." It says: "Your shared input is billed once per request no matter how many questions you evaluate (each question adds only its own instruction and criteria tokens)." There are "no output-token charges, variable thinking-token bills, or separate audio transcription and OCR costs."

Google says the Decisions API is "nearly 5x cheaper across the shared benchmark suite than Gemini 3.5 Flash-Lite with the Interactions API." VentureBeat reported a related figure from Google's shared tests: "Flash-Lite cost about 4.7 times as much as Decision 1 on that workload."

#Who can use it and when?

The API is available through the Gemini API in Google AI Studio. Google has not stated regional availability, paid-tier rate limits, or whether the endpoint is generally available. The blog's code sample uses the v1beta API version and the model ID gemini-decision-1.

VentureBeat reported that "Google lists free-tier limits of 200 requests and 50,000 input tokens per minute", and that "OpenRouter and Vercel AI Gateway also list the model." Google says ecosystem integrations "are rolling out across OpenRouter, LangChain, Vercel, LiteLLM, Pydantic AI, Pipecat, LiveKit, and OpenCode alongside the google-genai SDKs, with support coming soon to Gemini in Google Cloud."

#How to get it

Google's post tells developers to:

  1. Get an API key in Google AI Studio.
  2. Try the Doorstep Parcels, Speed Sketch and Find the Zoo Keeper demo apps.
  3. Follow the Python, JavaScript and REST guides in Google's Decisions API documentation.

#How does it compare?

Google says that on benchmarks spanning agent routing, reasoning, ranking and multimodal understanding, "Gemini Decision 1 leads across the suite, with the lowest calibration error." It also says the model keeps median latency at 50-100 ms of model compute and 150-250 ms end-to-end, "across multimodal inputs and contexts up to 16K tokens, remaining nearly flat from 1 to 100 questions per request."

Google's launch benchmarks, as reported by VentureBeat, show Decision 1 ahead on five of the seven tests run by all four models, at 92.2% on JevBench versus 90.9% for GPT-6 Luna and 90.0% for Jev 1.14. Jev led on Mind2Web Multimodal, 86.8% to 86.2%, and on When2Call, 82.0% to 79.9%. Against Gemini 3.5 Flash-Lite, Decision 1 scored 71.8% to 67.7% on the audio benchmark MMAU and 69.1% to 63.3% on Video-MME, with a calibration error of 0.0250, roughly half Jev's 0.0494. The shared tests cost $0.88 with Decision 1, $0.68 with Jev and $4.14 with Flash-Lite, VentureBeat reported.

VentureBeat reported that "The independent Artificial Analysis Decision Index v2.1 puts Gemini ninth of 18 at launch." In the scores it listed, Drex Max v1.1 was at 72.62, Jev 1.14 at 66.23 and Gemini Decision 1 at 65.98. On AA-Paperclip, Gemini "got every answer right on 33% of documents, versus 35.5% for Jev and 63% for Drex Max v1.1."

VentureBeat placed Google in a market that already includes TypeSafe's Jev, OpenAI's Decisions API on GPT-6 Luna, and Amazon's Strands Decider. It said OpenAI made its Decisions API generally available on October 22, 2026, and that OpenAI "charges $0.10 per million uncached input tokens and $0 for output, matching Google’s base rate." See OpenAI's Decisions API, which went generally available at $0.10 per 1M input tokens. VentureBeat reported Jev 1.14 at $0.042 per million input tokens with free output, Nace AI's Drex Max v1.1 at $0.03, and Cloudflare's Clef-Omni at $0.15 per million input tokens on Workers AI. It also reported that Amazon released Strands Decider 2B in October, "available under Apache 2.0, with serving code and a training recipe."

Google says its teams use the API: Gemini Notebook as a real-time router for everyday queries "with up to 40% lower latency", Stitch for playable wireframes, and CC from Google Labs to pick an emoji reaction from "a palette of more than 200 options".

See also Google's Agent Compute Engine and agent governance and Meta Muse Ember, a low-cost agent model in Preview on Google Cloud.

Our analysis: Google is entering a crowded decisions-API market that, according to VentureBeat, already includes OpenAI, TypeSafe and Amazon. Its own launch benchmarks, as reported by VentureBeat, put Decision 1 ahead on most tests Google ran, including calibration, while the independent index VentureBeat cited ranks it ninth of 18 and close to Jev. Teams planning to use the full context window or the 1,000-question limit should, as VentureBeat advised, "measure how long their own files and checklists take, including time spent uploading media."

#What is still unclear?

Google has not said whether the Decisions API is generally available, and it has not published regional availability or rate limits for paid tiers. Support for Gemini in Google Cloud is "coming soon", with no date. The free-tier limits and the Artificial Analysis figures come only from VentureBeat.