Introducing the Gemini Decisions API
On November 12 at 11:01 AM, Google revealed its new artificial intelligence tool, the Gemini Decisions API, designed to accelerate decision-making processes. This powerful API can handle up to 1000 simultaneous requests, processing a variety of data types—including text, images, audio, video, and PDFs—in just 150 to 250 milliseconds. Built on the Gemini Decision 1 model, it supports tasks with clear-cut conditions and can manage context lengths up to 256,000 tokens.
Capabilities and Pricing Details
The Gemini Decision 1 model evaluates predefined options by returning probabilities and confidence scores. It accommodates multiple question formats such as:
- Yes/no questions;
- Multiple-choice queries with up to 255 options;
- Rating tasks based on specific criteria.
Its processing speed ranges from 50 to 100 milliseconds per computation and remains consistent even when handling between 1 and 100 concurrent requests. In benchmark tests, Gemini Decision 1 outperforms competitors in tool selection, logical reasoning, ranking accuracy, and multimodal understanding, achieving the lowest calibration error among peers.
The service costs $0.10 per million input tokens, with no charges for output tokens. When developers submit shared data across multiple questions, they pay only once. The technology is already integrated into various Google products like Gemini Notebook, where it reduces request processing latency by up to 40%. Other applications include the Stitch API for creating interactive layouts, the CC assistant for emoji selection, and Google Search, where it functions as an additional security layer alongside Model Context Protocol (MCP) tools.
This launch arrives amid intensifying competition with OpenAI, which in October introduced its own Decisions API based on GPT-6 Luna. OpenAI’s solution processes decisions up to 10 times faster than its Responses API and can handle up to 400 questions per request, with pricing also starting at $0.10 per million input tokens. Google’s Gemini Decisions API is accessible through Google AI Studio, offering a free tier for development and testing. Integrations with OpenRouter, LangChain, Vercel, LiteLLM, and Pydantic AI are rolling out gradually.
Google's Gemini Decisions API launch highlights the fierce rivalry in the AI technology landscape as companies race to deliver faster and more efficient data processing solutions.
While OpenAI offers similar tools, Google's latest API aims to enhance user experience by minimizing processing delays. This development marks a significant advancement for developers and businesses seeking quick, reliable decision-making capabilities in their products and services.
As Google continues to innovate in the realm of artificial intelligence, its recent advancements in robotics also deserve attention. The Gemini Robotics 2.5 has demonstrated impressive accuracy during trials with Mercedes-Benz, showcasing the potential of AI to revolutionize manufacturing processes. This highlights the broader implications of AI technologies like the Gemini Decisions API in various sectors.