Check the model doing the work
Catch-up briefing: GitHub’s October 15 notice schedules three model retirements in Copilot for November 16. We checked it November 4. [1]
Copilot is GitHub’s coding assistant. A model is the AI doing the work; retirement means that choice will stop being offered there.
GitHub suggests Claude Opus 5.5 for Claude Opus 4.8 and Claude Opus 5, and GPT-6.1 Sol or GPT-6 Luna for GPT-5.6 Terra. [1]
Check the exception and team settings
The notice exempts individual Copilot subscribers on annual plans from the Opus 4.8 retirement. [1]
For Business and Enterprise, the suggested models are enabled under the default policy unless an administrator disabled that default or the model. Once enabled, they appear in Copilot Chat’s model list in supported apps. [1]
Available choices depend on your Copilot plan and where you use it. The current supported-model page does not independently repeat the November 16 schedule. [2]
Our take
Choose a replacement by a result you can check. Different models may handle the same job differently; a small, familiar task shows how much checking you need.
For a small shop’s website, use a copy and ask the replacement to correct one broken link. Write down the intended destination first, then inspect the changed file and check for unrelated edits.
Review proposed commands before approving them, and compare the time spent checking and correcting with your previous workflow. On a team account, if the replacement is missing, ask the administrator to check its model policy.
Keep the example and assign someone to review failures. We have not tested these models or changed a Copilot account.