GitHub Copilot Is Retiring 6 AI Models on December 14 — Here Are Their Replacements

Photo: Hippopx (public domain). Illustrative image.
What happened. GitHub announced on 13 November 2026 that six AI models will be deprecated across all Copilot experiences — Chat, inline edits, ask and agent modes, and code completions — on 14 December 2026. After that date, the old models disappear from the picker. For most users the switch is automatic, but the deadline is only a month away, and one model has no replacement named yet.
The full retirement table.
| Retiring model | Replacement |
|---|---|
| Claude Fable 5 | Claude Fable 5.2 |
| Claude Haiku 4.5 | Claude Haiku 5.5 |
| Gemini 3.8 Flash | Gemini 4 Hydrogen |
| Gemini 4 Argon | Gemini 4 Carbon |
| Grok 4.6 | Grok 4.7 |
| Kimi K3 | To be announced |
What you need to do. For Copilot Enterprise and Business customers, the replacements are automatically enabled under default model enablement — unless an admin has turned off the global default or disabled a model. In that case, nothing changes until the admin acts. No action is needed to remove the old models after 14 December; they are withdrawn by GitHub. Individual subscribers should check which model their plan defaults to, because the change also affects completions, not just chat. There is one exception to the clean sweep: Claude Haiku 4.5 stays available to individual Copilot subscribers on annual plans, so annual subscribers who rely on it are not forced off.
Why this round matters. The retirements read as a changing of the guard across every lab at once: Anthropic’s Fable and Haiku lines move a generation forward, Google’s Gemini 3.8 Flash gives way to Gemini 4 Hydrogen — the same model that went generally available in Copilot on 13 November, the day of the announcement — and xAI’s Grok steps from 4.6 to 4.7. The odd one out is Kimi K3, whose replacement is still “to be announced”, leaving Moonshot’s users without a named successor a month before the cut-off. Copilot’s model menu has become a rolling roster: models arrive, get superseded, and leave within months, which is good for capability and bad for anyone who has tuned a workflow around one specific model.
The wider context. The deprecation lands in a week of fast movement in coding agents: Google shipped Dynamic Workflows in Antigravity, letting teams run scripted multi-agent coding jobs, and our three-way comparison of Antigravity vs GitHub Copilot vs Claude Code weighs the three against each other — including Copilot’s pricing, which starts free and runs to $100 a month for Max. If you are choosing between the coding agents rather than just updating one, that comparison is the place to start; if you want the plain-English background, see What Is an AI Agent?.
Sources
- GitHub — “Upcoming deprecation of selected GitHub Copilot models in December”, GitHub Changelog (13 November 2026): https://github.blog/changelog/2026-11-13-upcoming-deprecation-of-selected-github-copilot-models-in-december
- FreshSynapse — “Google Antigravity Dynamic Workflows: Run a Team of Coding Agents From One Script”: https://freshsynapse.com/articles/antigravity-dynamic-workflows-coding-agents-2026-11-16/
- FreshSynapse — “Google Antigravity vs GitHub Copilot vs Claude Code: Which AI Coding Agent Should You Pick?”: https://freshsynapse.com/articles/antigravity-vs-copilot-vs-claude-code-2026-11-16/
- FreshSynapse — “Google Gemini 4 Hydrogen: Everyday Coding, Agentic Workflows and Consumer Tasks”: https://freshsynapse.com/articles/google-gemini-4-hydrogen-2026-11-13/