Microsoft's Decision-1 model enters the fast-growing AI decision model race
decoder ·
Microsoft has officially entered the rapidly expanding market for AI decision models with the launch of its new Decision-1 system. Built upon the Qwen3.5-9B architecture, this model is specifically optimized for rapid classification, evaluation, and routing tasks. According to Microsoft's internal testing across 36 benchmarks involving nearly 150,000 questions, Decision-1 achieves an accuracy rate of 83.5 percent. The company highlights that the model operates with a latency of just 85 milliseconds, making it 2.5 times faster than its closest competitor, H2O-Lightning-4B. This release positions Microsoft alongside other major players like OpenAI and Cloudflare in the race to provide fast, structured decision-making capabilities. While Jev pioneered the trend earlier this year, open-source small language models are quickly adapting and surpassing initial implementations. Decision-1 is currently accessible via Microsoft Foundry and OpenRouter, offering competitive pricing with free output tokens.
decoder