Microsoft releases Microsoft-Decision-1, a decision model built on Qwen3.5-9B

TL;DR

• On 9 October Microsoft released Microsoft-Decision-1, a model post-trained from Qwen3.5-9B that scores fixed answer options instead of writing text.
• In Microsoft's own 36-benchmark comparison it had the highest average accuracy, 83.5%, and the third-best calibration score, 92.2.
• It's in Microsoft Foundry and on OpenRouter at $0.042 per million input tokens, with free output and no open weights.