Microsoft releases decision scoring model Microsoft-Decision-1
Start here
Microsoft released the model Microsoft-Decision-1 for rapid decision scoring, now available on Microsoft Foundry and callable via OpenRouter. The model is post-trained from Qwen3.5-9B, outputs calibrated probability scores for fixed options, achieves the highest accuracy in comparisons across 36 benchmarks, and has a P50 latency about 35 times faster than GPT-6 Sol. Across 5,250 requests on 11 benchmarks, it refuses harmful behavior while maintaining utility, with an average decision flip rate of 1.3% under 8 perturbations.
Generated by AI from reports · Updated 18 hr ago
Progress timeline
- 公众号Doonsec· Discussion微软推出决策AI模型：企业智能决策赛道迎来新玩家(opens the original post in a new tab)
- 网页开源中国 · 资讯· Discussion微软发布基于 Qwen 后训练的决策模型 “Microsoft-Decision-1”(opens the original post in a new tab)
- Command LineMicrosoft 发布 Microsoft-Decision-1：面向快速决策评分的模型
Microsoft 发布 Microsoft-Decision-1 快速决策评分模型，已在 Microsoft Foundry 上线并可通过 OpenRouter 调用。该模型由 Qwen3.5-9B 后训练而来，为固定选项输出校准概率分数，在 36 项基准比较中准确率最高，P50 延迟约为 GPT-6 Sol 的 35 倍快。在 11 项基准的 5,250 条请求中它拒绝有害行为并保持效用，8 种扰动下平均决策翻转率 1.3%。
- X@OpenRouter· Discussion微软发布决策模型 Microsoft-Decision-1，基于 Qwen3.5-9B（OpenRouter）(opens the original post in a new tab)
- X@satyanadella· Discussion微软发布决策模型 Microsoft-Decision-1，基于 Qwen3.5-9B(opens the original post in a new tab)
Related discussion
- 网页TechmemeMicrosoft unveils Microsoft-Decision-1, a fast decision-scoring model trained on Qwen3.5-9B, and says it will soon rebase it on MAI, OpenAI, and other models (Achint Srivastava/Command Line)(opens the original post in a new tab)
- 公众号Doonsec微软推出决策AI模型：企业智能决策赛道迎来新玩家(opens the original post in a new tab)
- 网页开源中国 · 资讯微软发布基于 Qwen 后训练的决策模型 “Microsoft-Decision-1”(opens the original post in a new tab)
- 网页IT之家微软推出 Microsoft-Decision-1 决策模型，基于 Qwen3.5-9B(opens the original post in a new tab)
- X@OpenRouter微软发布决策模型 Microsoft-Decision-1，基于 Qwen3.5-9B（OpenRouter）(opens the original post in a new tab)
- X@satyanadella微软发布决策模型 Microsoft-Decision-1，基于 Qwen3.5-9B(opens the original post in a new tab)
Attention trend
New sources per day
- Oct 10New5 sources (3 outlets,2 accounts)
- Oct 11New2 sources (2 outlets,0 accounts)
- Oct 12New0 sources (0 outlets,0 accounts)