Microsoft Shows Seven New MAI Products and Bets on Price
At Ignite, Microsoft presented seven new MAI products. According to the company, its coding model MAI Code 2 gives 98% of the quality of GPT-6.1 Sol at 40% of the cost.
At Ignite, Microsoft's yearly conference for business customers, the company showed seven new products from MAI, its in-house AI lab. They include models for writing code, making images, handling speech and spotting security threats, plus a new chip called Maia 200. Microsoft's main argument is price. By its numbers, the coding model MAI Code 2 delivers 98% of the quality of GPT-6.1 Sol, an OpenAI model, at 40% of the cost.
AI agents are programs that write code on their own, often around the clock. That work is billed in tokens, the small pieces of text a model reads and writes. Because those bills add up, companies now weigh the cost of a finished task, not just a model's place in the rankings. People who already use GitHub Copilot, Microsoft's coding assistant, can pick MAI Code 2 there today as a cheaper option.
Microsoft ran every comparison itself, on tests it chose. It also says the models are not yet tuned for hard work such as scientific research or very long chains of agent steps. It compares MAI Code 2 with Opus 5.5, Anthropic's top model, and claims 96% of its results at 5% of the price. That puts the pitch on price, while the quality benchmark is still someone else's model. Nobody outside Microsoft has checked these figures yet, so teams should test the models on their own work.
Source: microsoft.ai
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides

Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles