GitHub announced on June 17, 2026 that auto mode in Copilot Chat is now generally available for all users on github.com and the GitHub mobile app. In the official changelog, GitHub says auto mode chooses a model based on request complexity and current availability, and paid users get a 10% discount while using it.

That sounds like a convenience feature, but for QA engineers it is really a test repeatability and governance change. If Copilot can route the same task to different models over time, teams should verify whether summaries, code suggestions, bug classifications, and triage recommendations stay stable enough for real workflows.

What changed on June 17, 2026

  • Auto mode in Copilot Chat is now generally available for all Copilot plans on github.com and the GitHub mobile app.
  • GitHub says auto mode selects a model based on task complexity and real-time availability.
  • GitHub also says users can still inspect which model answered by hovering over the response.
  • Paid subscribers get a 10% discount on model costs while using auto mode.

GitHub’s docs add an important detail: the models used by auto mode can change over time, subject to subscription plan and admin policies. The docs also say auto model selection is available more broadly across Copilot products, but the June 17 changelog item specifically covers Copilot Chat on github.com and the mobile app.

Why this matters for QA engineers

Many QA teams now use AI assistants for flaky-test triage, exploratory test ideas, bug-report cleanup, log summarization, and pull-request review. Auto routing can help reduce friction, but it also introduces a practical question: will the same prompt produce consistently usable output next week if the underlying model changes?

  • If you use Copilot to classify CI failures, compare whether auto mode keeps the same root-cause ranking across repeated runs.
  • If you use Copilot for test generation, check whether locator choices and assertion quality stay consistent enough for review.
  • If your organization restricts models by policy, confirm auto mode respects those rules in the contexts your testers actually use.
  • If cost reporting matters, validate whether the 10% discount changes your default prompt-routing strategy for routine QA tasks.

Three checks worth running this week

  1. Repeatability check: Run the same triage or test-generation prompt multiple times and log which model responded plus how much the output changed.
  2. Policy check: Ask an admin or enterprise owner to confirm that allowed-model policies and exclusions behave as expected with auto mode enabled.
  3. Cost versus quality check: Compare a few common QA prompts in auto mode versus a pinned model to see whether the cheaper path is still good enough for production review work.

Why this matters for QA engineers

Auto model selection can be useful for QA teams if it lowers rate limits, reduces friction, and keeps response quality high enough for everyday review work. But it should be treated like any other test-tooling change: verify output stability, document which workflows are safe to automate, and keep evidence of which model answered when results affect release decisions.

Sources