On 15 October 2026 GitHub announced that three models will be removed from all Copilot experiences on 16 November 2026: Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra. That covers Copilot Chat, inline edits, ask and agent modes, and code completions.
What replaces them
| Retiring model | GitHub’s suggested alternative |
|---|---|
| Claude Opus 4.8 | Claude Opus 5.5 |
| Claude Opus 5 | Claude Opus 5.5 |
| GPT-5.6 Terra | GPT-6.1 Sol or GPT-6 Luna |
The one exception: individual Copilot subscribers on annual plans keep access to Opus 4.8. For Business and Enterprise customers, the suggested alternatives are switched on automatically under default model enablement, unless an administrator has turned off the global default or disabled the model. GitHub’s guidance to users is to move workflows and integrations to supported models before the date.
The pattern behind the notice
This is not a one-off. GitHub has published model retirement notices repeatedly this year, and we covered the same problem when GPT-5.5 left Codex on 14 October. The models your developers use on a project can change underneath it several times before launch.
What it means for a buyer
Switching model is usually painless for a developer typing prompts. It is much less painless when a model name is hard-coded into your product, a build pipeline, or a prompt library that was tuned for one model’s habits. A cheerful “no action required” from the vendor covers the vendor’s side, not your tests.
So what
When you commission AI-assisted development or an AI feature, ask three things: which models does the supplier depend on today, where is the model name configured (one setting, or scattered through the code), and who re-tests the product when a model is swapped. A good answer is one setting, an automated test suite, and a named person responsible. If you are planning an AI-powered product or want a supplier that treats model changes as routine maintenance, see our AI-assisted development approach or start a project.