Describe the bug
When switching models via /model with an effort level specified, the effort is not applied. The status bar retains the previous model's effort level. The change notification correctly reflects the requested effort, so the issue is in application, not display.
Note: Running /model and picking the same model with a different effort level updates the effort level correctly.
I've attached screenshots sequentially stepping through the /model command behavior.
The statusline effort and chat message do not mention the same effort value.
Baseline. GPT-5.3-Codex · high(3%). No issue here; this is the correct starting state.
Switched to gpt-5.2-codex with low effort. Notification says low was applied.
Status bar shows high.
Behavior: Effort not applied.
Switched to gpt-5.3-codex with medium effort. Notification says medium was applied.
However, the statusline shows low.
Behavior: Effort not applied.
Effort explicitly changed for gpt-5.3-codex from low to high.
Because we didn't do a model swap, this worked as intended.
Behavior: No issue.
Switched from gpt-5.3-codex (high) to gpt-5.4-mini (low). Statusline still says high.
Behavior: Effort not applied
Affected version
1.0.42
Steps to reproduce the behavior
- Start a session. Note current model and effort level in the status bar (e.g. GPT-5.3-Codex · high).
- Run /model and switch to a different model with a different effort level (e.g. gpt-5.4-mini at low).
- Observe the notification: "Model changed from gpt-5.3-codex (high) to gpt-5.4-mini (low)".
- Observe the status bar — it still shows high, not low.
Expected behavior
The effort level selected in /model should be applied during the switch. The status bar should reflect it.
Additional context
Reproduced across multiple model switches (gpt-5.2-codex, gpt-5.3-codex, gpt-5.4-mini) in a single session.
Tested on both a Windows 11 and a Mac installation of Github Copilot CLI; Encountered identical behavior.
I did not encounter this issue before 1.0.41
Describe the bug
When switching models via /model with an effort level specified, the effort is not applied. The status bar retains the previous model's effort level. The change notification correctly reflects the requested effort, so the issue is in application, not display.
Note: Running /model and picking the same model with a different effort level updates the effort level correctly.
I've attached screenshots sequentially stepping through the
/modelcommand behavior.The statusline effort and chat message do not mention the same effort value.
Affected version
1.0.42
Steps to reproduce the behavior
Expected behavior
The effort level selected in /model should be applied during the switch. The status bar should reflect it.
Additional context
Reproduced across multiple model switches (gpt-5.2-codex, gpt-5.3-codex, gpt-5.4-mini) in a single session.
Tested on both a Windows 11 and a Mac installation of Github Copilot CLI; Encountered identical behavior.
I did not encounter this issue before 1.0.41