Microsoft announced its new AI models, MAI Code 2 and Code 2 Flash, at the keynote address of Ignite 2026. Designed for everyday coding tasks, MAI Code 2 is touted as offering high performance at lower cost, while Code 2 Flash is said to enable use at an even lower price.
Microsoft unveiled MAI Code 2 and Code 2 Flash on 17 November local time at its annual Microsoft Ignite 2026 conference as new frontier AI models focused on everyday coding tasks. The announcement also reflected a strategic aim to reduce AI costs and reliance on external providers, with an emphasis on integration with Microsoft's own product lineup.
The company claims that MAI Code 2 delivers performance equivalent to Anthropic's Claude Opus 5.5 at approximately 5% of the price, and said it could provide performance comparable to OpenAI's GPT-6.1 Sol and Opus 5.5 at lower cost. Code 2 Flash aims to deliver performance equivalent to GPT-6 Luna and Claude Haiku 5.5 at a substantially lower cost.
However, the figures presented are based on usage-based API pricing and differ from monthly service fees, while the benchmarks were supplied by Microsoft itself. The figures could change if independent third-party verification is conducted in the future.
Microsoft also announced new features derived from Code 2 and Code 2 Flash, including MAI Work Flash. These are expected to support a broad range of everyday business tasks, such as generating formulas in Excel, creating documents in Word and tracking action items in Teams.