[Bug] All models show as 'Blocked/Disabled' after June 16 outage

[yzeng58]

Description

After the GitHub Copilot outage on June 16, 2026 (17:45-18:15 UTC), the Copilot CLI model selection interface shows all available models as "Blocked / Disabled", preventing users from selecting any model or starting new sessions.

Expected Behavior

The model selection interface should show the available models based on the user's organization policy (as visible on https://github.com/settings/copilot/features).

Actual Behavior

• All models are marked as "Blocked / Disabled" in the CLI model selection UI
• The message states: "These models are not currently available. They may be disabled by your organization's policy or network issues."
• However, the GitHub web interface (https://github.com/settings/copilot/features) shows most models as available and enabled
• Cannot start new Copilot CLI sessions or select models
• Remote debugging connection also fails

Steps to Reproduce

1. After the June 16, 2026 outage (resolved around 18:15 UTC)
2. Run gh copilot or try to select a model in an active session
3. Observe that all models show as "Blocked / Disabled"

Environment

• Copilot CLI Version: 1.0.64-0
• OS: macOS (Darwin)
• Time of Issue: June 16, 2026 ~20:30 PDT (June 17, 2026 03:30 UTC)
• GitHub Account: Enterprise account with Microsoft organization
• Organization Policy: Models are enabled per web settings page

Additional Context

• This issue appeared immediately after the upstream provider outage mentioned in GitHub status (Opus 4.8 degradation)
• Pre-existing Copilot CLI sessions that were started before the outage continue to work with their originally selected model
• The GitHub web UI correctly shows available models (Claude Sonnet 4.5, Opus 4.8, Gemini, GPT-5 series, etc.)
• gh api /user/copilot/entitlements returns HTTP 404, suggesting the API endpoint may still have issues

Expected Fix

The CLI should either:

1. Gracefully handle API failures and use cached model availability data
2. Properly refresh model availability after service restoration
3. Provide a manual cache clear/refresh command for users to recover from such incidents

Workaround Attempted

Tried to clear CLI cache but issue persists (possibly needs server-side fix or API recovery).

[grantborthwick]

This reproduces for me on both public and ghe accounts.

[tomfong]

Same for me. All high-end GPT and Claude models suddenly unavailable.

[allexkonov]

same here, Max plan

[allexkonov]

the issue is resolved, I see all the proper models now

[dereklegenzoff]

Apologies for the inconvenience here folks. This incident is now resolved. More details here: https://www.githubstatus.com/incidents/rfmjwng33vjf