|
Google has introduced a new artificial intelligence (AI) tool that can evaluate up to 1,000 questions across text, images, audio, video and PDF documents in a single pass, as the company expands its AI offerings for faster and more cost-efficient decision-making by businesses and developers.
The Gemini Decisions API, powered by the new Gemini Decision 1 model, is designed for tasks such as selecting tools, checking policies, routing queries to AI models and evaluating information. Unlike conventional generative AI systems that produce responses token by token, the model scores available options in a single processing step and returns probabilities along with confidence scores. In a blog post, Google said the API can deliver end-to-end response times of 150-250 milliseconds, with pricing set at USD 0.10 per million input tokens and no charges for output tokens. The company said the shared input is billed only once, regardless of the number of questions evaluated in a request. The API supports three types of decisions: yes-or-no assessments, multiple-choice selections and scoring against an ordered set of criteria. Each answer includes a confidence score, allowing developers to direct uncertain cases to a fallback system or human reviewer. The model can process up to 256,000 tokens of multimodal input, allowing users to combine different formats in a single request. Google said Gemini Decision 1 performed strongly across benchmarks covering reasoning, ranking, agent routing and multimodal understanding, recording the lowest calibration error among the models compared. It also said the API was nearly five times cheaper than Gemini 3.5 Flash-Lite with the Interactions API across the shared benchmark suite. The company said its teams are already using the technology in Gemini Notebook to route queries to suitable AI models, in Stitch for faster product design and in Google Search as a safety layer for detecting adversarial content. The launch follows Google's expansion of access to its Gemini 4 Carbon model earlier this week. Carbon is designed for complex coding, cybersecurity, financial research and other tasks requiring extended reasoning and coordination between multiple AI agents. While Carbon focuses on handling complex, long-running tasks, the Decisions API is aimed at making individual evaluations faster and less expensive, broadening Google's AI tools for developers and businesses. (ANI)
|
