Signal Microsoft unveils MAI-Code-1-Flash to lessen reliance on OpenAI

Summary

Microsoft unveiled MAI-Code-1-Flash, its own coding AI model, at the Build 2026 conference in San Francisco on 2 June 2026, one of seven new MAI models from Mustafa Suleyman's AI Superintelligence Team. The 5-billion-parameter model was trained directly on GitHub Copilot's production workflows and reportedly outperforms Claude Haiku 4.5 by 16 percentage points on SWE-Bench Pro while using 60% fewer tokens on complex tasks. It is priced at $0.75 per million input tokens ($0.075 cached) and $4.50 per million output tokens under Copilot's AI Credits system, and is available to GitHub Copilot Free, Student, Pro, Pro+ and Max personal plans, with Business/Enterprise access not yet timelined. The launch follows the April 2026 end of Microsoft's seven-year exclusive partnership with OpenAI, with CEO Satya Nadella describing it as a shift from “consuming a frontier model to fully participating at the frontier.” Independent analysis (Simon Willison) and the model's own data card noted that OpenAI models were used in data preparation, complicating the “fully in-house” framing.