Microsoft Launches Decision-1 Model on Qwen3.5-9B

Microsoft introduced the Decision-1 model on October 9, 2026. The system is built on top of Alibaba's Qwen3.5-9B model. It launched in public preview on Microsoft Foundry. OpenRouter confirmed the model went live under the identifier microsoft/microsoft-decision-1 within roughly a day of the announcement. The model accepts text inputs up to 32,000 tokens. It does not generate explanations or handle images, audio, or video.

Microsoft claims Decision-1 runs 35 times faster than GPT-6 Sol. It runs 4.5 times faster than Quyet-1.0-Large. It was reported to be 2.5 times faster than H2O-Lightning-4B v1.1. The pricing is $0.042 per million input tokens with free output. The model was tested on approximately 150,000 held-out questions.

Satya Nadella introduced the model on X. He stated it is being tested for incident response, quality control, and scientific discovery. Xbox Research used Decision-1 to categorize more than 10,000 pieces of open-ended feedback into predefined themes. Xbox Research reported that using Decision-1 for this task was 80–100 times faster than alternatives.