Skip to content

GitHub Copilot retires Claude Opus 4.8, Opus 5 and GPT-5.6 Terra

GitHub deprecated three models across most Copilot experiences on November 16, 2026, naming replacements and leaving one fallback path open.

Draftship Editorial3 min read

What happened on November 16

GitHub has deprecated three models across most GitHub Copilot experiences, effective November 16, 2026. The GitHub Copilot model deprecations cover Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra, and they apply to Copilot Chat, inline edits, ask and agent modes, and code completions. The changes were first announced on October 15, 2026, according to GitHub's changelog entry.

GitHub paired each retired model with a suggested replacement:

Deprecated model Deprecation date Suggested alternative
Claude Opus 4.8 2026-11-16 Claude Opus 5.5
Claude Opus 5 2026-11-16 Claude Opus 5.5
GPT-5.6 Terra 2026-11-16 GPT-6.1 Sol or GPT-6 Luna

The exceptions GitHub carved out

The deprecation is not uniform. Claude Opus 4.8 remains available to individual GitHub Copilot subscribers on annual plans. Claude Opus 5 has been removed from the model picker but stays reachable for runtime refusal fallback through existing authorized access. Existing enterprise model policies continue to apply on top of both arrangements.

Those two carve-outs matter because they mean some sessions will keep resolving to a model that is otherwise retired. A team that assumes Opus 4.8 is gone everywhere, or that Opus 5 is unreachable in every code path, would be working from a wrong picture.

What admins and developers have to do

For Copilot Enterprise and Copilot Business customers, the suggested alternatives are automatically enabled under default model enablement, unless an administrator has turned off the global default or explicitly disabled the model. If the global default was turned off, access can be restored through the alternative models' model policies in Copilot settings. Once enabled, users see the model in the Copilot Chat model selector in supported experiences.

GitHub states that no action is required to remove the deprecated models themselves. The work is in the other direction: making sure something supported is switched on, and that anything pinned to a retired model name is updated. GitHub's guidance is to update workflows and integrations to use supported models. Enterprise customers with concerns are pointed to their account manager.

Limits and open questions

The announcement says "most" Copilot experiences but does not list which ones fall outside that set. It does not say whether API access, third-party integrations or bring-your-own-key setups are affected, nor does it address pricing differences between the retired models and their replacements. Whether Claude Opus 5.5, GPT-6.1 Sol and GPT-6 Luna are available on every plan tier is not stated either. A separate changelog entry dated November 13 flags an upcoming deprecation of selected Copilot models in December, but that post's contents are not included here, so the scope of the next round is unknown.

What this means for people building with AI

Hardcoded model names age badly. This round retired three models at once and split their replacements across three different successors, which suggests any prompt, script or agent configuration that names a model explicitly should be treated as a maintenance item rather than a constant. If you are choosing a model for AI features inside an app you are building, the same discipline applies: check which models a platform currently offers before wiring one in, as covered in Draftship's documentation on using the AI chat and choosing a model. Draftship is not affiliated with GitHub, and nothing here indicates Draftship is affected by these Copilot changes.

Primary sources

This article was drafted with AI assistance from the primary sources listed below and checked automatically against them. An editor hasn't reviewed it yet.

AI coding agents

opencode v2.3.2 fixes GPT-6.2 prompt caching billing

opencode v2.3.2 sends the system prompt as a leading developer message on every Responses route so GPT-6.2 caches it, cutting uncached input per step from about 14,000 tokens to 3 on one Copilot model.