AI Guides
GitHub Copilot Model Deprecations: November 2026 Migration Checklist
Updated October 26, 2026. GitHub has scheduled another GitHub Copilot model retirement for November 16, 2026. If you or your team rely on Claude Opus 4.8, Claude Opus 5, or GPT-5.6 Terra in Copilot, review the change now and test the suggested alternatives before the date. This guide summarizes GitHub’s official notice and gives you a practical migration checklist.
Which GitHub Copilot models are changing on November 16?
| Model being deprecated | Suggested alternative | Important note |
|---|---|---|
| Claude Opus 4.8 | Claude Opus 5.5 | GitHub says this retirement does not apply to individual Copilot subscribers on annual plans, who will retain an Opus offering. |
| Claude Opus 5 | Claude Opus 5.5 | Check that the replacement is available in your Copilot plan and client. |
| GPT-5.6 Terra | GPT-6.1 Sol or GPT-6 Luna | Choose based on task quality and availability, then test your usual workflow. |
GitHub’s notice lists November 16, 2026 as the deprecation date across Copilot Chat, inline edits, ask and agent modes, and code completions. The Claude Opus 4.8 exception for individual annual subscribers is explicit, so do not assume every account will see the same change. Read the official November deprecation notice before changing a production workflow.
What about the models retired on October 19?
GitHub already deprecated six other model options on October 19, 2026: Gemini 3.7 Flash, GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5 mini, and Grok 4.5. GitHub suggested Gemini 3.8 Flash, GPT-6.1 Sol, GPT-6 Luna, and Grok 4.7 as alternatives for those entries. If you are updating a team checklist, include both the October 19 changes and the upcoming November 16 retirement. The two notices cover different model sets.
See the October 19 retirement notice and the live GitHub Copilot AI model documentation for the current supported-model list.
Who needs to take action?
- Individual users: check your model picker in the Copilot client you actually use, such as VS Code or GitHub.com. Availability can differ by plan and client.
- Copilot Business and Enterprise administrators: review model policies and global defaults. GitHub says suggested alternatives are automatically enabled under default model enablement unless an administrator has turned off the global default or explicitly disabled the model.
- Teams with pinned workflows: review instructions, team documentation, saved prompts, and any automation that assumes a specific model is available.
- Users on annual Copilot plans: note the specific Claude Opus 4.8 exception in GitHub’s notice; verify your own account rather than assuming the model will disappear on November 16.
GitHub says no action is required to remove the deprecated models after they are retired. That does not mean you should ignore the change: you still need to choose a supported model if a workflow depends on one of the retiring options.
Migration checklist before November 16
- Inventory model use. Check Copilot Chat, inline edits, ask mode, agent mode, code completion, team instructions, and any IDE-specific settings. Record which people and workflows depend on each affected model.
- Check real availability. Open the model picker in each supported client and verify the alternative is available under your plan and organization policy. Do not rely only on a screenshot or an old internal guide.
- Review administrator policies. For Business or Enterprise, confirm whether the global default is enabled and whether specific models have been restricted. If needed, update model policies before asking users to switch.
- Choose a replacement by task. Start with GitHub’s suggested alternative, then compare available options for the actual work: routine edits, multi-file refactors, tests, debugging, documentation, or agent-driven tasks.
- Run a small regression set. Use representative repository tasks and the same instructions. Check code correctness, test results, patch size, tool use, latency, and how much manual correction is needed.
- Check request usage and plan limits. Model availability and premium-request multipliers can vary. Review GitHub’s current model comparison and plan documentation rather than assuming the replacement has the same usage cost.
- Update team guidance. Replace obsolete model names in onboarding notes, prompt templates, runbooks, and troubleshooting pages. Tell teammates what to select if the old model no longer appears.
- Recheck after the change. On or before November 16, verify the model picker and test the workflows again. Keep a short record of the selected model and any behavior differences.
How to choose between the suggested replacements
Do not treat the suggested alternatives as a universal quality ranking. GitHub’s recommendation is a practical starting point, but the best choice depends on your subscription, client, task, and policy settings. Test the smallest set of models that are actually available to you.
| Workload | Evaluation focus |
|---|---|
| Small edits and code completion | Correctness, latency, and whether suggestions fit the surrounding code. |
| Multi-file refactors | Test pass rate, completeness, unintended changes, and review time. |
| Debugging and complex reasoning | Quality of the diagnosis, reproducibility, and whether the proposed fix addresses the root cause. |
| Agent mode and tool use | Task completion, permission boundaries, tool-call accuracy, and safe recovery from failures. |
| Documentation and explanations | Technical accuracy, clarity, and consistency with the repository. |
For a fair comparison, use the same repository, issue, tests, instructions, and acceptance criteria. Record not only whether the model produces a plausible patch, but whether the patch passes tests and can be reviewed safely. For agentic coding, keep destructive commands, deployments, and external changes behind appropriate approval and permission controls.
Does this retire the same models from their providers’ APIs?
No. This notice describes model availability within GitHub Copilot experiences. It should not be interpreted as a blanket announcement that the corresponding provider API has been discontinued. If your application calls a model directly through a provider API, check that provider’s own deprecation schedule separately.
Related Aiverio resources
- Best AI Coding Assistants — compare coding assistants for different development workflows.
- GitHub Copilot vs. Claude Code — compare two coding workflows before choosing a broader tool.
- Cursor vs. Claude Code — explore alternatives for repository-level coding tasks.
- AI Browser Agent Security Checklist — review permission and safety controls for agentic workflows.
Frequently asked questions
When are Claude Opus 4.8, Claude Opus 5, and GPT-5.6 Terra deprecated in GitHub Copilot?
GitHub lists November 16, 2026. Claude Opus 4.8 remains available to individual GitHub Copilot subscribers on annual plans under the exception in the official notice.
Will GitHub automatically switch my selected model?
The notice explains how suggested alternatives are enabled under default model policies, but it does not promise that every individual workflow will be automatically remapped. Check the model picker and your organization’s policies.
Do I need to delete the retired model from a configuration?
GitHub says no action is required to remove deprecated models after retirement. You should still update your workflow or team guidance if it depends on a model that will no longer be selectable.
Where can I verify supported models?
Use GitHub’s current AI models documentation and the dated November 16 announcement. Recheck both before making account or policy changes.
