Microsoft has released Microsoft-Decision-1, a decision-scoring model built on Alibaba's Qwen3.5-9B. Both sources describe it as returning calibrated probabilities for fixed answer options rather than generating open-ended text, a design aimed at reducing hallucination and cost for tasks like routing, classification, verification, and agent control.
The Register places the release in a crowded field: Jev, OpenAI's Decisions API, Cloudflare's Clef, and more than 100 other decision models are now competing. Microsoft claims its model is 2.5x faster than H2O-Lightning-4B and 2.8x faster than Jev in latency tests, with 83.5% accuracy across 36 benchmarks and a price of $0.042 per million input tokens (output tokens free). These figures are vendor claims and not independently verified.
The two sources differ in scope. MarkTechPost's brief announcement focuses on the model's mechanics and use cases, while The Register adds competitive context and notes that Microsoft plans to rebase Decision-1 on its own models and those from OpenAI in the future. That rebasing detail does not appear in the shorter source.