Microsoft’s first advanced reasoning AI is here
Microsoft’s first advanced reasoning AI is here
微软首款高级推理 AI 正式发布
MAI-Thinking-1 is one of seven new models announced at Build 2026. MAI-Thinking-1 是微软在 Build 2026 大会上发布的七款新模型之一。
Microsoft announced a bunch of new in-house AI models at Build 2026, including a new “flagship” model: MAI-Thinking-1. It’s an ambitious step into model development for Microsoft, which introduced its initial in-house models last year — before then, it had relied on OpenAI’s models. The two companies recently renegotiated their deal to loosen ties. 微软在 Build 2026 大会上发布了一系列全新的自研 AI 模型，其中包括一款新的“旗舰”模型：MAI-Thinking-1。对于微软而言，这是其在模型开发领域迈出的雄心勃勃的一步。微软去年才推出了首批自研模型，在此之前，该公司一直依赖 OpenAI 的模型。两家公司最近重新谈判了协议，以放宽彼此间的合作绑定。
According to Microsoft, MAI-Thinking-1 is a “medium-sized model” that “matches leading models” on “key” software engineering benchmarks. Microsoft says the company “trained it from the ground up on clean data, without distillation from third-party models.” 据微软介绍，MAI-Thinking-1 是一款“中型模型”，在“关键”软件工程基准测试中“表现与领先模型相当”。微软表示，该公司“完全基于干净的数据从零开始训练了该模型，没有使用任何第三方模型的蒸馏技术。”
As for other models announced today, they’re focused on image generation, transcription, voice, and coding. 至于今天发布的其他模型，它们主要专注于图像生成、转录、语音和编程领域。
MAI-Image 2.5 and the flash version can do text-to-image and image editing. MAI-Transcribe-1.5 is “five times faster than competing models.” MAI-Voice-2 and the flash version of that model (which Microsoft says is “coming soon”) add 15 new languages and new options for voices. The new coding model, MAI-Code-1-Flash, is “inference-efficient” and is integrated into GitHub Copilot and Visual Studio Code. MAI-Image 2.5 及其闪电版（flash version）支持文本生成图像和图像编辑功能。MAI-Transcribe-1.5 的转录速度“比同类竞争模型快五倍”。MAI-Voice-2 及其闪电版（微软称该版本“即将推出”）增加了 15 种新语言和新的语音选项。全新的编程模型 MAI-Code-1-Flash 具备“推理高效”的特点，并已集成到 GitHub Copilot 和 Visual Studio Code 中。