Describe the bug

When both  copilot_cli_opus_medium_effort_default  and  copilot_cli_gpt_5_4_mini_for_explore  feature flags are active (assigned server-side), the CLI throws the following error repeatedly during sub-agent execution:

Execution failed: Error: Reasoning effort 'medium' is not supported for model 'claude-haiku-4.5'.

The  explore  and  task  built-in agents are defined with  model: claude-haiku-4.5  in  definitions/explore.agent.yaml  and  definitions/task.agent.yaml . The  copilot_cli_opus_medium_effort_default  flag applies  reasoning_effort: medium  globally — including to these sub-agents — before the  copilot_cli_gpt_5_4_mini_for_explore  swap has a chance to replace the model. As a result,  claude-haiku-4.5  receives a  reasoning_effort  value it does not support.

Affected version

1.0.77

Steps to reproduce the behavior

1. Be assigned both flags:  copilot_cli_opus_medium_effort_default: true  and  copilot_cli_gpt_5_4_mini_for_explore: true  (visible in  config.json  under  expAssignmentsCache )
2. Trigger any task that uses the  explore  or  task  sub-agent
3. Observe the error in output

Expected behavior

Either the model swap happens before the reasoning effort is applied, or  reasoning_effort  is omitted for models that do not support it (as already fixed in main: "Allow reasoning effort to be omitted so the server can select the default").

Additional context

Platform:
Windows ( win32-x64 )