On November 13, 2026, GitHub Copilot’s GPT models showed degraded availability — GPT-6.1 Terra, GPT-5.6 Sol, and GPT-6 Luna returned elevated error rates in Copilot code completion and chat. GitHub resolved the incident at 21:27 UTC the same day after an upstream model provider implemented a mitigation. There is no user-side fix — but there is a one-click workaround GitHub itself recommends.
Quick Fix Summary
- Problem: On November 13, 2026, GitHub Copilot’s GPT models showed degraded availability — GPT-6.1 Terra, GPT-5.6 Sol, and GPT-6 Luna returned elevated error rates in Copilot code completion and chat.
- Fix: None user-side. GitHub resolved the incident at 21:27 UTC on November 13 after an upstream model provider implemented a mitigation. If it happens again, switch to another model or select ‘Auto’ in the model picker and keep coding.
- What happened: GitHub status incident “Copilot experiencing degraded model availability for some GPT models” — investigating 20:13 UTC, resolved 21:27 UTC (1 hour 14 minutes). Cause: an issue with an upstream model provider.
Technical Root Cause
| Error Code | Affected Version | Root Cause | Current Status |
|---|---|---|---|
| No error code — elevated error rates on GPT models | GitHub Copilot (code completion + chat, VS Code and other IDE surfaces); models gpt-5.6-sol-fast, gpt-5.6-sol, gpt-6.1-terra, gpt-6-luna | Issue with an upstream model provider (named by GitHub, not specified further); provider implemented a mitigation | Resolved 21:27 UTC, November 13, 2026 |
Step-by-Step: What to Do When Copilot’s Models Fail
Step 1 — Check GitHub’s Status Page First
Open https://www.githubstatus.com and look at the Copilot AI Model Providers component. On November 13 it showed degraded availability for the GPT models. If the status page shows a Copilot incident, the problem is upstream — nothing in your editor, extension, or network is causing it. This is the same first move as our Fix: ChatGPT Work Mode HTTP 500 Error — when the vendor’s own status page shows an incident, stop troubleshooting your own setup.
Step 2 — Switch to Auto or Another Model
In the Copilot Chat model picker, select ‘Auto’ or pick a model that is not listed as affected. That is GitHub’s own recommendation from the incident timeline — the degraded models were GPT-6.1 Terra, GPT-5.6 Sol, and GPT-6 Luna, so anything else in the picker keeps working. Keep coding; your completions and chats just run on a different model until the affected ones recover.
Step 3 — Do Not Reinstall VS Code or the Extension
Reinstalling VS Code or the Copilot extension changes nothing here — the model list and the errors both come from GitHub’s servers. If the models are degraded, a fresh install shows the same warnings. Save the reinstall for a problem that is actually on your machine, like the sign-in failure we covered in Fix: Claude Code Failed to Refresh OAuth Token — that one is local, this one is not.
Step 4 — Retry the Affected Model Only After Resolution
After the status page marks the incident resolved (November 13: 21:27 UTC), switch back to your usual model and confirm completions work. If it still errors after resolution, that is a new problem — check the model picker warnings and try ‘Auto’ once more before digging into your setup.
Step 5 — Know the Pattern
This was the third Copilot model-availability incident in five days — November 9 (GPT-5.3 Codex, GPT-5.6 Luna/Sol, 49 minutes), November 11 (multi-product, 33 minutes), and November 13 (this one, 1 hour 14 minutes). Same upstream-provider cause each time. If Copilot’s GPT models go wobbly again, your fastest move is always the same: status page first, switch to ‘Auto’, keep coding.
While Copilot’s GPT models are down, keep a second AI assistant ready
If your coding flow depends on one AI product, a degraded-model incident stops you cold. Claude Pro runs the same kind of coding chats — completions, explanations, and long sessions — so you can keep shipping while GitHub fixes its side. (Affiliate link — DebugRadar earns a commission if you sign up through it.)
FAQ
Is GitHub Copilot down right now?
The November 13 incident is resolved — GitHub marked it resolved at 21:27 UTC after an upstream provider implemented a mitigation. If Copilot’s models fail again, check githubstatus.com first: if the Copilot AI Model Providers component is degraded, the problem is upstream, not your setup.
Which Copilot models were affected?
GPT-6.1 Terra, GPT-5.6 Sol (including the fast variant gpt-5.6-sol-fast), and GPT-6 Luna. Code completion and chat using those models showed elevated error rates, and GitHub’s in-product model warnings activated for them.
What should I do while Copilot’s models are degraded?
Switch to another model or select ‘Auto’ in the model picker — that is GitHub’s own recommendation — and keep coding. Do not reinstall VS Code or the Copilot extension; the model list comes from GitHub’s servers. Retry the affected model only after the status page shows resolved.
Sources: GitHub status page (githubstatus.com), November 13, 2026; Pingoru GitHub Copilot timeline; PulsAPI status report, November 14, 2026.
Verified by DebugRadar Engineering — Tested on Production Environments
DebugRadar’s engineering team tracks this issue against the vendor’s official status page and reproduces the reported failure before publishing. If the vendor resolves it, we update the guide — check the status links above for the latest state.