Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

reuters.com.axios.com.axios.com.Microsoft on Tuesday, Nov. 17, introduced a set of in-house AI models for coding, cybersecurity, image generation and voice at its Ignite conference in San Francisco. The company is trying to win customers from OpenAI and Anthropic by offering similar performance for much less money, [Reuters](https://www.reuters.com/legal/litigation/microsoft-unveils-ai-models-coding-cybersecurity-it-takes-anthropic-openai-2026-11-17/) Thomson Reuters Corporation reported. [Axios](https://www.axios.com/2026/11/17/microsoft-ai-models-openai-anthropic-ignite) counted six new models. Microsoft's own [announcement](https://microsoft.ai/news/microsoft-ignite-2026/) called it a family of seven.axios.com+2
The release shows Microsoft relying less on outside partners for advanced AI. Until now it has largely depended on OpenAI and other companies for its most capable models, [Axios](https://www.axios.com/2026/11/17/microsoft-ai-models-openai-anthropic-ignite) noted.axios.com
MAI Code 2 is the main release. Microsoft averaged its scores across four coding benchmarks, including DeepSWE 1.1 and Terminal Bench 2.1. On that measure, it says MAI Code 2 delivers 98% of the performance of OpenAI's GPT-6.1 Sol at about 40% of the price, and 96% of Anthropic's Claude Opus 5.5 at 5% of the price, [according to Microsoft AI](https://microsoft.ai/news/introducing-mai-code-2-and-mai-code-2-flash/). The smaller MAI Code 2 Flash is built for routine edits and fixes. Microsoft says it performs about as well as GPT-6 Luna at 65% of the cost and Claude Haiku 5.5 at 20%. The company also said the models are "explicitly not yet tuned" for harder work such as basic science research.microsoft.ai
Both coding models are now in GitHub Copilot. MAI Code 2 is offered on Pro, Pro+, Max, Business and Enterprise plans, while Flash also reaches Free and Student users, [GitHub's changelog](https://github.blog/changelog/2026-11-17-mai-code-2-and-mai-code-2-flash-in-github-copilot/) said. Code 2 Flash is also in a limited preview on Microsoft Foundry, according to [Microsoft's Foundry blog](https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/from-model-choice-to-your-durable-advantage-build-ai-that-keeps-improving/4573609).github.blog+1
MAI Cyber 2 Flash is a security model. Microsoft says it matches GPT-6 Astra on the CyberGym benchmark for less than 2% of the cost per task, according to [Microsoft AI](https://microsoft.ai/news/microsoft-ignite-2026/). Only a limited group of customers will get it, [Reuters](https://www.reuters.com/legal/litigation/microsoft-unveils-ai-models-coding-cybersecurity-it-takes-anthropic-openai-2026-11-17/) reported. The lineup also includes MAI Image 2.7 plus new transcription and real-time voice models, [Axios](https://www.axios.com/2026/11/17/microsoft-ai-models-openai-anthropic-ignite) said.microsoft.ai+2
Microsoft AI chief Mustafa Suleyman told Axios the company is making progress toward its own general-purpose frontier model. He cited state-of-the-art pre-training results but gave no release timeline. "My team is building true superintelligence," he said, [per Axios](https://www.axios.com/2026/11/17/microsoft-ai-models-openai-anthropic-ignite). Microsoft will also license model weights to enterprise customers so they can customize the models, as long as they deploy them inside Azure.axios.com
Speaking to Reuters, Suleyman said building models for specific domains keeps costs down. "This isn't just one model. It's not a flash in the pan," he said, [according to Reuters](https://www.reuters.com/legal/litigation/microsoft-unveils-ai-models-coding-cybersecurity-it-takes-anthropic-openai-2026-11-17/). He also called alignment, meaning making sure AI behaves as intended, his top priority.reuters.com
The benchmarks Microsoft chose for its comparisons are selective, and some of its price advantages come from promotional discounts, [Axios](https://www.axios.com/2026/11/17/microsoft-ai-models-openai-anthropic-ignite) noted. On the same day, OpenAI announced GPT-6.2 Astra and GPT-6.2 Sol and priced the new Astra 20% below the previous version.axios.com