微软用自研数据+强化学习打造了强推理模型，做 AI 推理或模型训练的团队值得关注其“爬山机器”方法论，尤其是 35B 激活参数就能达到接近顶尖水平的效率。
微软发布了 MAI-Thinking-1，这是其自研推理模型系列的首个成果。该模型采用 1T 总参数的混合专家架构，每次推理仅激活 35B 参数，在 AIME 2025 上达到 97.0%，LiveCodeBench v6 上 87.7%，SWE-Bench Pro 上 52.8%。微软称其训练流程为“爬山机器”，通过持续优化数据、训练、奖励和安全测试形成闭环。预训练基于 30T 主要人工生成 token，避免使用第三方模型蒸馏，随后通过强化学习提升数学、编程、工具使用和安全能力。这标志着微软在推理模型领域建立了完整的自研能力。
Microsoft unveiled MAI-Thinking-1. So Microsoft n…
Microsoft unveiled MAI-Thinking-1.
So Microsoft now has a full in-house pipeline for building stronger reasoning models again and again.
Microsoft calls this system a “hill-climbing machine,” meaning it keeps improving the data, training setup, rewards, safety tests, and evaluations as one connected process.
Strong for its size, including 97.0% on AIME 2025, 87.7% on LiveCodeBench v6, and 52.8% on SWE-Bench Pro.
MAI-Thinking-1 is the first model from that process, using 35B active parameters inside a 1T total parameter mixture-of-experts model, where only part of the model runs for each token.
The base model was trained from scratch on 30T mostly human-generated tokens, with Microsoft saying it avoided third-party model distillation during pre-training.
After that, the team used reinforcement learning, which means the model practiced tasks and improved from feedback, to teach math reasoning, coding, tool use, helpfulness, and safety.