Google has rolled out the Gemini Decisions API, an advanced artificial intelligence tool aimed at enabling businesses and developers to make quicker and more cost-effective decisions. This new API, powered by the Gemini Decision 1 model, is capable of evaluating up to 1,000 questions in a single request, spanning text, images, audio, video, and PDF documents.
Unlike traditional AI models that process responses incrementally, the Gemini Decisions API evaluates multiple options simultaneously, offering probability estimates and confidence scores for decision-making. This feature is particularly useful for tasks such as selecting tools, checking policies, routing queries, and analyzing information.
Google claims the API delivers response times between 150 and 250 milliseconds, with pricing beginning at $0.10 per million input tokens. Notably, there are no charges for output tokens, and shared input incurs a single billing, regardless of the number of questions processed. The API supports a variety of decision formats, including yes-or-no answers, multiple-choice selections, and ranking options based on specified criteria.
With its multimodal capabilities, the API can handle up to 256,000 tokens of combined input, making it apt for complex business applications. Google has already incorporated this technology into several of its products, including Gemini Notebook, Stitch, and Google Search, to enhance query routing, product design, and content safety.
This launch marks a significant expansion of Google’s AI ecosystem, offering developers and businesses a more efficient and affordable solution for handling individual evaluations and automated decision-making tasks.