Microsoft Launches MAI-Code-1-Flash and MAI-Thinking-1 at Build 2026

At Build this week (June 2-3), Microsoft unveiled its first proprietary AI models built entirely in-house without OpenAI data. MAI-Code-1-Flash is a 5B-parameter coding model that outperforms Claude Haiku 4.5 across all core coding benchmarks (+16 points on SWE-Bench Pro: 51.2% vs 35.2%) and solves harder problems with 60% fewer tokens — it's rolling out to GitHub Copilot users in VS Code now. MAI-Thinking-1 is a 35B active-parameter MoE reasoning model with a 256K context window that scores 97% on AIME 2025 and 53% on SWE-Bench Pro (matching Claude Opus 4.6) — in private preview via Microsoft Foundry. For engineers: these are direct Copilot integrations, which means competitive pressure on Anthropic and OpenAI pricing is only going up.

---

Sources: Microsoft AI · Neowin · CNBC