GitHub Copilot will remove six older models on October 19, and names each replacement

Why it mattersA team pinning Copilot to a named model in code review, agent mode or chat has one month to switch or accept the automatic swap, and the automatic swap sends the two GPT-5.4 variants to a different model family.
GitHub will retire six models across every GitHub Copilot experience on October 19, 2026, according to a September 18 changelog post. The affected experiences named in the post are Copilot Chat, inline edits, ask and agent modes, and code completions.
The list
The six models being removed, and the alternative GitHub is naming for each:
- Gemini 3.7 Flash to Gemini 3.8 Flash
- GPT-5.5 to GPT-5.6 Sol
- GPT-5.4 to GPT-5.6 Sol
- GPT-5.4 mini to GPT-5.6 Luna
- GPT-5 mini to GPT-5.6 Luna
- Grok 4.5 to Grok 4.6
Two of the retirements are inside the same GPT family, but GPT-5.4 and GPT-5.4 mini are each being redirected across model tiers rather than to their nearest same-tier sibling: the two mini variants collapse into one alternative (Luna), and the standard GPT-5.4 and GPT-5.5 both point at Sol.
What happens if you do nothing
GitHub says that under default model enablement the suggested alternatives are turned on automatically for Copilot Enterprise and Copilot Business customers, unless an administrator has turned off the global default or explicitly disabled the alternative model. Teams that have turned off the global default have to enable the alternative through the model policies in Copilot settings before October 19, or that alternative will not be reachable.
The company also says no action is required to remove the models after they are deprecated. GitHub will remove them from the picker on that date.
Why the same date matters twice
This is the third Copilot deprecation announcement in six weeks. GitHub retired an earlier batch on September 1, then announced on September 3 that Gemini 3.5, Gemini 3.6, Kimi K2-7 and Claude Opus 4.7 would go on October 2, then this. So the October 19 removal takes effect two weeks after the October 2 removal, and the pace matters more than any single line on the list: a Copilot workflow that names a specific model in a script, an eval, a company policy document or a persisted user setting is now on a rolling deadline, not a one-off patch.
Teams that grade code review or agent output against a named model, or pin the model in an eval that runs on every pull request, have to write down which of the two GPT-5.6 variants replaces each of their retired GPT-5.4 pins, since GitHub is now pointing full and mini calls at different destinations.
Source
Source: GitHub Changelog
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.


