GitHub will retire three AI models in its coding assistant on November 16, 2026. Developers and organizations using them must check recommended replacements and access settings. They must update workflows and integrations to supported models. An exception keeps one retiring model available to individual annual subscribers.
1. Retirement covers chat through code completion
GitHub published the change notice on October 15, 2026. The retirement applies across GitHub Copilot. It covers Copilot Chat, inline edits made directly in code, ask mode, agent mode for delegated tasks, and code completion.
The notice does not mean model choices are changing only in a particular chat interface. Using models while editing or for code completion is subject to the same retirement schedule.
2. Replacement models and the annual subscription exception
The recommended replacement for Claude Opus 4.8 is Claude Opus 5.5. However, the existing model will remain available to individual annual subscribers. GitHub said this exception preserves their access to an Opus model.
The recommended replacement for Claude Opus 5 is also Claude Opus 5.5. The annual subscription exception described above applies only to the retiring model in the preceding paragraph, not to this model.
GPT-5.6 Terra will also be retired on the same date.
The recommended alternatives are GPT-6.1 Sol or GPT-6 Luna.
3. Automatic enablement is not the same as migration
For Copilot Enterprise and Copilot Business customers, recommended replacements are enabled automatically based on default model settings. This makes them available; it does not mean model selections in existing workflows and integrations are automatically changed.
Automatic enablement does not apply if an administrator has turned off the global default or explicitly disabled a model. If the global default is off, access to replacements can be enabled through model policies in Copilot settings.
Once enabled, the models appear in the chat model picker in supported environments.
GitHub asked users to update workflows and integrations to supported models before the retirement date.
4. Distinguish removal steps from undisclosed information
GitHub said users need not take any additional action to remove the models after retirement. This guidance concerns the removal process.
The original notice provides no performance comparisons, measurement conditions, or cost changes for the recommended replacements. A recommendation alone does not establish better speed or quality, or lower spending.
A separate supported models document was provided for the full list of available models.
GitHub Enterprise customers can contact their account team with questions or concerns.
The key is to distinguish replacement model availability from actual use in workflows, and to check the annual subscription exception.
