Simon Willison's WeblogJun 2, 2026, 10:21 PMimportant 72

Microsoft's new MAI models

Microsoft announced two compact MAI text models for reasoning and coding, emphasizing performance, cost, and licensed training data.

Microsoft announced MAI-Thinking-1, a 35B reasoning model available to select early partners, and MAI-Code-1-Flash, a 5B coding model rolling out to GitHub Copilot individual users in VS Code. Simon Willison highlights their relatively small parameter counts and Microsoft's claim that MAI-Thinking-1 was preferred to Sonnet 4.6 in internal blind evaluations. He also questions what Microsoft's clean and appropriately licensed training data claims mean in practice.

Microsoft released two new large text language models. MAI-Thinking-1 is a 35B-parameter reasoning model, currently available only to certain early partners. MAI-Code-1-Flash is a 5B-parameter code-development model, built for GitHub Copilot and VS Code usage scenarios, with the goal of reducing cost while maintaining performance, and it has begun gradually rolling out to individual GitHub Copilot users within VS Code.

Full summary

Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.

See Pro plans →

Want the original English / full article?

Read on Simon Willison's Weblog →

Summaries are AI-generated; the original article is authoritative.