GitHub Copilot drops GPT-4 Turbo for Microsoft's own model — what Project Polaris means for teams standardised on Copilot

Microsoft is switching GitHub Copilot's default engine from GPT-4 Turbo to Project Polaris, its own in-house coding model, in an automatic rollout this month — a reminder that 'which AI coding tool' and 'which model underneath it' are now two separate decisions.

GitHub Copilot’s default engine changes this month. Project Polaris, Microsoft’s own mixture-of-experts coding model, replaces GPT-4 Turbo as the out-of-the-box default for every Copilot subscriber, migrated automatically. Enterprise customers get an optional three-month fallback to GPT-4 Turbo through November if they need to validate the switch before committing; everyone else is moved over with no action required and no extra cost.

Microsoft is framing this as a performance upgrade, and it’s also a strategic one: it cuts Copilot’s dependency on OpenAI’s models for its default experience, something Microsoft has been building toward since it started letting Copilot users pick Claude and Gemini models manually. Polaris being the default — not just an option — is the bigger signal. Most developers don’t change defaults.

That matters for anyone whose team standardised on Copilot as “the AI coding tool” without much thought to which model sits behind it. The model doing the reasoning is changing under a name that hasn’t. Benchmarks Microsoft has published claim Polaris beats GPT-4 Turbo on standard coding tasks, but those numbers haven’t been independently verified yet, and coding-benchmark claims from any vendor are worth treating with some scepticism until they hold up in your own codebase.

So what

If your team runs on Copilot, budget a sanity check in the next few weeks: run your usual review workflow against a few real diffs and see whether output quality holds, especially on anything security- or domain-sensitive. This is also a useful moment to revisit the bigger question — Copilot, Cursor, and Claude Code all now let you choose or default to different underlying models, so “which AI coding tool” and “which model” are genuinely separate decisions, and the right answer for a regulated healthcare build isn’t necessarily the right answer for a fast-moving MVP. If you want a team that tracks this shift for you rather than defaulting to whatever ships in the box, see how we approach AI-assisted development or get in touch to talk through your stack.

← All research

More research.

GitHub's Copilot impact dashboard now counts chat and VS Code agent use, and drops the 'passive users' label

On 6 November 2026 GitHub added Copilot Chat and the VS Code Agents window to its Copilot impact dashboard, bringing it to nine tracked experiences, and renamed 'Passive users' to 'Chat-first users'.

Read ›

Claude Code 2.1.319: a setting that denies permission prompts so AI agent runs don't stall

Claude Code 2.1.319 (6 November 2026) adds a permissions.denyInsteadOfPrompting setting and moves long MCP tool calls to the background, a sign AI-assisted development is being built for unattended runs.

Read ›

GitHub now reports how often developers act on Copilot code review comments

On 5 November 2026 GitHub added code review metrics, including an addressed rate, to the Copilot usage metrics API. For AI software development buyers, it is the first first-party number for how much AI review feedback is acted on.

Read ›

Dependabot can now keep AI agent skills and prompts up to date

GitHub's Dependabot gained Agent Package Manager (APM) support on 3 November 2026, treating AI agent skills, prompts and instructions as dependencies that need updating like any other software.

Read ›

GitHub Copilot is now inside Xcode's own coding assistant: what iOS app owners should ask

GitHub put Copilot into Xcode's built-in coding assistant on 4 November 2026, in public preview with Xcode 27.3 beta. AI-assisted iOS app development is now a built-in choice of agent, not a plug-in.

Read ›

GitHub's Copilot usage API now separates bring-your-own-key and custom models: AI spend visibility for software buyers

GitHub's Copilot usage metrics API now reports bring-your-own-key (BYOK) and enterprise custom model usage separately, as of 3 November 2026. Teams using AI coding tools can see more of where usage goes.

Read ›

Start a project

Building with AI? Let's talk.

Tell us what you are building. We'll give you an honest view of the right approach.

Start a projectSee our work