Microsoft has introduced Decision-1, an AI model designed for structured decision-making tasks such as classification, routing, and prioritization. Unlike general-purpose models, it selects answers from a fixed list rather than generating text. The model operates by assigning calibrated probability scores to options, allowing businesses to gauge confidence levels for automated workflows.
Built on the open-weight Qwen3.5-9B model from Alibaba, Decision-1 features a single-pass architecture that processes up to 32,000 tokens at once. Microsoft reports that the model is 4.5 times faster than its nearest competitor, Quyet-1.0-Large, and achieved the highest accuracy in internal benchmarks involving nearly 150,000 questions.
The model is currently undergoing internal testing at Microsoft for incident response, quality control, and scientific discovery. External users can access Decision-1 via Microsoft Foundry, with a planned deployment on OpenRouter. Future iterations of the model are expected to be based on Microsoft AI and OpenAI architectures.
