Copilot loses the whole GPT model family for half an hour

Sources: GitHub model-provider incident, GitHub Actions delay incident, GitHub Actions failure incident, OpenAI status
Summary: GitHub logged a major incident from 09:42 to 10:11 UTC on 2026-07-25 in which GPT-5.2, GPT-5.3-Codex, GPT-5.4, GPT-5.4 Mini, GPT-5.6 Sol, GPT-5.6 Terra, and GPT-5.6 Luna were all degraded across Copilot products and IDE surfaces. GitHub attributed the degradation to an issue with an upstream model provider and said a root cause analysis will follow. OpenAI opened an elevated-error-rates incident at 09:17 UTC the same morning, applied a mitigation at 10:02 UTC, and marked it resolved at 11:08 UTC, then opened a second elevated-error-rates incident from 11:35 to 11:57 UTC. Neither OpenAI entry lists affected components, so the overlap with the Copilot degradation is not confirmed to be the same fault. GitHub logged two Actions incidents the same day. A minor one from 08:59 to 09:25 UTC delayed workflow run starts. A critical one from 12:31 to 13:13 UTC caused workflow run failures and delays, was mitigated at 12:58 UTC, and left jobs queued before 12:40 UTC needing a retry.
Why it matters: One upstream fault took out every GPT option in the Copilot picker at once, so switching models inside a single vendor's list is not provider redundancy.
Follow-up: Watch for GitHub's root cause analyses on both the model-provider and the Actions incidents, and for whether OpenAI publishes the scope of the 2026-07-25 incidents.