THE CRUNCH
Microsoft has joined the decision model race with Decision-1, a small model built for fast, structured decisions such as classifications, evaluations and routing. According to Microsoft, it has the potential to guide and control agents through complex environments. Decision-1 is based on Qwen3.5-9B, and Microsoft says it was the most accurate model tested across 36 benchmarks covering nearly 150,000 questions, while running 2.5 times faster than the runner-up, H2O-Lightning-4B.
The category is barely a month old: Jev introduced the decision model concept in mid-September, and Cloudflare released its open-source Clef and Clef-flash models on Workers AI in early October, describing them as producing bounded structured outputs cheaply, quickly and consistently, with Clef then leading the Jev Decision Index.
In Microsoft's benchmark comparison, Decision-1 tops the table with 83.5% accuracy and 85 ms latency, ahead of Jev 1.13.0. Cloudflare's Clef models, which are also based on Qwen, were not included in the comparison, so the claimed lead covers only the models Microsoft tested.
Decision-1 is available through Microsoft Foundry and OpenRouter. Input tokens cost $0.042 per million, and output tokens are free. The pressure on Jev is palpable: the idea caught on quickly, but the technology was rapidly adapted and surpassed using open small language models.