Describe the bug
Summary
Issue: Models enabled at the organization level are not appearing in the Copilot CLI's /model command.
Details:
- I am an admin of the organization
- I enabled GPT-5.6 and Claude Sonnet 5 in
Copilot → Models → Policies on GitHub
- I disabled Claude Sonnet 4.6
- Waited 20+ minutes for propagation
- Cleared cache:
rm -rf ~/.copilot/cache ~/.copilot/state
- Restarted CLI with
/restart
- Both GPT-5.6 and Sonnet 4.6 do not appear in
/model output
- CLI version: 1.0.77
Expected: All models enabled at the org level should appear in /model within a reasonable time
Actual: Only older models are listed; newly enabled models remain absent
Suggestion: Provide a /refresh-models command or similar to force a refresh of the available models list without waiting for server propagation.
Current available models (from /model):
- Auto, Claude Sonnet 4.6, Claude Haiku 4.5, Claude Opus 4.7
- GPT-5.5, GPT-5.3-Codex, GPT-5 mini
- (and older claude variants)
Missing: GPT-5.6 (and any other newly enabled models)
Affected version
1.0.77
Steps to reproduce the behavior
- As an organization admin, go to GitHub Organization → Settings → Copilot → Models
- Enable a new model (e.g., GPT-5.6) that is not currently in use across the organization
- Save the changes
- Wait 15-20+ minutes for server propagation
- On your local machine, run the Copilot CLI:
- Expected: The newly enabled model (GPT-5.6) should appear in the list
- Actual: The model does not appear; only previously available models are shown
Additional Troubleshooting Attempted
• Ran /restart in the CLI
• Cleared cache: rm -rf ~/.copilot/cache ~/.copilot/state
• Logged out and back in via gh auth
• Updated to CLI version 1.0.77 (latest)
• Waited 20+ minutes (far exceeding typical propagation time)
Result: Model list remains unchanged; newly enabled models do not appear.
Expected behavior
All models enabled at the org level should appear in /model within a reasonable time
Additional context
- ZSH 5.9 (arm64-apple-darwin24.0)
- MacBook Pro Sequoia 15.7.8
Describe the bug
Summary
Issue: Models enabled at the organization level are not appearing in the Copilot CLI's
/modelcommand.Details:
Copilot → Models → Policieson GitHubrm -rf ~/.copilot/cache ~/.copilot/state/restart/modeloutputExpected: All models enabled at the org level should appear in
/modelwithin a reasonable timeActual: Only older models are listed; newly enabled models remain absent
Suggestion: Provide a
/refresh-modelscommand or similar to force a refresh of the available models list without waiting for server propagation.Current available models (from
/model):Missing: GPT-5.6 (and any other newly enabled models)
Affected version
1.0.77
Steps to reproduce the behavior
Additional Troubleshooting Attempted
• Ran /restart in the CLI
• Cleared cache: rm -rf ~/.copilot/cache ~/.copilot/state
• Logged out and back in via gh auth
• Updated to CLI version 1.0.77 (latest)
• Waited 20+ minutes (far exceeding typical propagation time)
Result: Model list remains unchanged; newly enabled models do not appear.
Expected behavior
All models enabled at the org level should appear in /model within a reasonable time
Additional context