September 5, 2026: GitHub has expanded enterprise-managed settings so organizations can choose their preferred Copilot model for new conversations and assign different defaults to enterprise teams. The September 2 announcement says the feature is generally available for Copilot Business and Enterprise in the Copilot app, Copilot CLI and Visual Studio Code. Source: GitHub’s release announcement.
What changed
Administrators can set an enterprise default and specialize it by team membership. GitHub describes marking the model key as overridable and configuring team files through team-mappings.json. Users without a team-specific value inherit the enterprise default.
The setting controls the starting model, but users can still choose a different model for an individual conversation. GitHub’s reference supports automatic selection or a specific model and version, and recommends the top-level model key for new configurations. Source: enterprise-managed settings reference.
Why this matters for QA engineers
QATechTools analysis: team defaults could make AI-assisted test generation and defect triage easier to compare by giving a group a consistent starting point. They do not guarantee reproducible answers: testers should record the actual model, prompt, repository revision and tool permissions alongside each evaluation.
What to validate before a wider rollout
Start a new conversation in each supported client with a mapped-team account and an account using the enterprise fallback. Confirm the initial model, then check that changing it for one conversation behaves as expected. Keep a small fixed set of test-generation and failure-triage tasks to compare output quality after changing the default.
For the QA pilot, evaluate whether generated assertions catch seeded defects and whether proposed fixes preserve existing coverage. Treat those results as your own evidence; GitHub’s settings announcement does not establish that any model is better for your test suite.