GitHub Copilot's Default Model Changes This Month: Project Polaris Replaces GPT-4 Turbo for 4.7M+ Subscribers

GitHub Copilot switches its default reasoning engine to Project Polaris — Microsoft's in-house MoE coding model — replacing GPT-4 Turbo for 4.7M+ paid subscribers starting this month.

Project Polaris uses a mixture-of-experts architecture with language-specific sub-modules, announced at Microsoft Build 2026 (June 2) and rolling out as the Copilot default in August. The model handles multi-file refactoring via chain-of-thought reasoning at inference time. All subscribers get a three-month window to opt back to GPT-4 Turbo. If your team's Copilot policies, .devcontainer configs, or API integrations pin gpt-4-turbo by name, audit those now — the silent migration is underway.

Why it matters: Copilot now runs on a proprietary, non-API-accessible model. For teams weighing Claude Code against Copilot for agentic workflows, the comparison shifted: you're benchmarking against a black-box engine you can't call from your own pipelines.

Sources: Project Polaris Set as Default — Windows Forum · Microsoft Build 2026: Polaris Recap — Enterprise DNA · GitHub Copilot Replaces GPT-4 With Project Polaris — TechTimes