Microsoft introduced two new AI coding models, MAI Code 2 and MAI Code 2 Flash, at Ignite 2026, claiming performance close to leading rivals while costing significantly less.
At the Ignite 2026 conference, Microsoft announced MAI Code 2 and MAI Code 2 Flash, its first in‑house AI models for software development. The company says the models are already integrated into GitHub Copilot and can be selected through the platform’s model picker in Visual Studio, VS Code and the Copilot website.
Microsoft positions MAI Code 2 as the more capable offering, targeting complex tasks such as multi‑file refactoring, deep debugging and full‑repository analysis. In internal benchmarks the model achieved 98% of GPT‑6.1 Sol’s performance while costing only 40% of the latter, and 96% of Anthropic’s Claude Opus 5.5 at 5% of its price.

The lighter MAI Code 2 Flash focuses on speed for routine work like code explanations, minor bug fixes and targeted edits. Microsoft claims it matches GPT‑6 Luna and Claude Haiku 5.5 in performance while using 65% and 20% of their respective costs.
Access to the models depends on Copilot subscription level. MAI Code 2 is available to Pro, Pro+, Max, Business and Enterprise plans, whereas Code 2 Flash is also offered to Free and Student users within usage limits. Both models are part of Microsoft’s broader push to expand its proprietary AI portfolio beyond coding.
TechTimes Commentary
TechTimes notes that Microsoft’s pricing claims could make its AI coding tools more attractive to a broad range of developers, especially those already using GitHub Copilot. If the cost advantage holds up in real‑world use, it may pressure OpenAI and Anthropic to adjust their pricing structures for comparable services. The rollout also signals Microsoft’s intent to reduce reliance on external AI providers and to build a self‑sufficient ecosystem, a trend that could reshape the competitive landscape for enterprise development tools.







