Microsoft 在 Ignite 发布 7 个 MAI 模型,覆盖编程、安全与图像
Microsoft 一次发了 7 个自研模型,重点看两个:MAI Code 2 编程能力贴着 Claude Opus 5.5 但只要 1/20 的价格,MAI Cyber 2 Flash 安全测试打平 GPT-6 Astra 成本不到 2%。做编程和安全类应用的开发者可以先算一笔账。
Microsoft 在 Ignite 大会上一次性发布了 7 个 MAI 系列模型,覆盖编程、网络防御、图像编辑、实时语音等场景。其中 MAI Code 2 在编程基准测试上与 Claude Opus 5.5 相差几分,但价格仅为 Anthropic 的二十分之一。
MAI Image 2.7 登上 Artificial Analysis 图像编辑榜单第一名,主打图像编辑能力的可视化质量。MAI Cyber 2 Flash 在一项网络安全测试中与 GPT-6 Astra 打平,而单任务成本不到后者的 2%。
对开发者来说,这批模型的核心卖点是价格:MAI Code 2 以约二十分之一的成本逼近 Anthropic 旗舰编程模型的水平,MAI Cyber 2 Flash 也以极低成本对标 OpenAI 的安全测试成绩。这类"成绩接近头部、价格低一至两个数量级"的组合,直接冲击 Anthropic 和 OpenAI 在编程与安全两个垂直场景的定价空间。
SHIPPING: Microsoft introduced seven MAI models at its Ignite event, including new systems for coding, cyber defense, image editing, live voice, and more. Highlights include: MAI Code 2: within a few points of Claude Opus 5.5 on coding benchmarks, for a twentieth of what Anthropic charges. MAI Image 2.7: moves to No. 1 on Artificial Analysis’ image editing board. MAI Cyber 2 Flash: ties GPT-6 Astra on a cybersecurity test, costing under 2% as much per task.
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Microsoft, MAI, MAI Code 2, Claude Opus 5.5.
Source attribution: curated from The Rundown AI (@therundownai). Cross-source corroboration: 11 independent sources have covered this development (including TestingCatalog, Artificial Analysis, Decoder), confirming industry-wide signal strength.