Agent Upstart

Cool things people are building with agents — and practical help getting the most out of them.

Playbooks · Coding tasks

Re-pin Copilot models before retirements

Written November 2, 2026

Claude Opus 4.8, Claude Opus 5, and GPT-5.6 Terra retire November 16. Check what you pinned, and map each one to its replacement before it stops.

What retires on November 16

GitHub's changelog entry of October 15, 2026 says the following models will be deprecated across all GitHub Copilot experiences — Copilot Chat, inline edits, ask and agent modes, and code completions — on November 16, 2026:

One exception applies, and only to one model: Claude Opus 4.8 will remain available to individual GitHub Copilot subscribers on annual plans so they retain an Opus offering, and the deprecation of Claude Opus 4.8 does not apply to those customers. The exception does not cover Opus 5 or GPT-5.6 Terra, and it does not cover Business or Enterprise seats. GitHub asks users to update workflows and integrations to use the supported models before the date, and says no action is required to remove the models after they have been deprecated.

Before you start

Changing the model for Copilot Chat does not change the model for inline suggestions — they are separate settings. Check both. If you are on a Copilot Business subscription, your organization must grant members the ability to switch to a different model, and a Copilot extension may override the model you select. If you use Copilot Fusion or automatic model selection instead of a pinned model, Fusion selects the model itself; this page covers pinned models, so check the picker in your client for what Fusion offers rather than assuming the table above applies to it.

Steps

  1. List every pinned model. In each client you use, open the Copilot Chat model picker and write down the model selected. In Visual Studio Code, open Copilot Chat from the chat icon in the title bar and use the model dropdown at the bottom of the chat view; in JetBrains IDEs, use the status bar icon, open GitHub Copilot Chat, and use the model dropdown in the chat view. GitHub's documentation gives the picker steps for each client, including Visual Studio, Eclipse, and Xcode. Then search your workflows, integrations, scripts, and settings for the three retiring model names — GitHub's changelog asks for workflows and integrations to be updated, and a name pinned outside the picker is easy to miss.
  2. Check each one against the retirement table. Mark every occurrence of Claude Opus 4.8, Claude Opus 5, or GPT-5.6 Terra. For Opus 4.8, check first whether the annual-plan exception applies to you: it covers individual subscribers on annual plans only. Everyone else moves Opus 4.8, and everyone moves Opus 5 and GPT-5.6 Terra.
  3. Re-pin each one to its suggested alternative. Map Opus 4.8 and Opus 5 to Claude Opus 5.5, and GPT-5.6 Terra to GPT-6.1 Sol or GPT-6 Luna. For Terra, the choice between the two alternatives is yours to test in step 4 — GitHub names both, and neither is presented here as the better one. If a replacement does not appear in your picker, check with your administrator: under default model enablement, GitHub says the suggested alternatives are automatically enabled for Copilot Enterprise and Copilot Business customers unless an administrator turned off the global default or explicitly disabled the model, in which case access is enabled through the model policies in Copilot settings.
  4. Test one task on each replacement. Run a task you already ran on the retiring model, on the replacement, and compare the result. A re-pin that has not run your actual work is not finished. Keep the old model's output for the comparison; a replacement that answers differently on your task is a finding, not a failure of the test.
  5. Check code completions separately. Re-pinning Chat does not change the model used for inline suggestions. Open the completions model setting in your client and check it against the same retirement table. If it names a retiring model, change it too, and test one completion on the replacement.

What can go wrong

Limits

Related playbooks

Sources

Back to: Coding playbooks hub · Also: Copilot Fusion is generally available: what it does to your AI-credit bill · Decide whether Copilot's assisted permissions should approve anything for you · Review a pull request with a coding agent