Microsoft launches decision model with Qwen3.5-9B foundation

On October 9, Microsoft announced the release of its new decision model, Microsoft-Decision-1, built on the Qwen3.5-9B foundation. The model is designed for structured decision tasks, offering top-tier performance in response latency and decision quality, surpassing large language models and traditional decision systems. It is now available on Microsoft Foundry and integrated with OpenRouter. In a comprehensive benchmark test involving 36 evaluations and over 150,000 isolated test cases, Microsoft-Decision-1 achieved the highest accuracy. It also outperformed Quyet-1.0-Large by 4.5 times and GPT-6 Sol by 35 times in response speed. Microsoft conducted a security evaluation using 11 benchmarks and 5,250 requests, confirming its effectiveness in blocking malicious activities without compromising usability. Pricing is set at $0.042 per million input tokens, with output free.