openclaw - 💡(How to fix) Fix Control UI model switcher prepends active provider to new model ID on provider change [1 comments, 2 participants]

The model switcher dropdown in the Control UI incorrectly constructs the model identifier when switching providers. Instead of using the selected model's fully-qualified provider/model format from the configured model list, it prepends the currently active provider to the new model's ID. So if you're on openai/gpt-4o and select Claude from the dropdown, it sends openai/claude-sonnet-4-6 to the gateway instead of anthropic/claude-sonnet-4-6, resulting in a 'model not allowed' error. Workaround is to use the command directly.

Fix Plan

The fix involves updating the model switcher dropdown logic to use the fully-qualified provider/model format.

• Update the dropdown selection handler to extract the selected model's fully-qualified ID from the configured model list.
• Use the extracted ID to construct the request to the gateway.

Example code snippet:

# Assuming 'models' is a list of dictionaries with 'provider' and 'model' keys
selected_model = models[dropdown_selected_index]
fully_qualified_id = f"{selected_model['provider']}/{selected_model['model']}"

# Use the fully_qualified_id to construct the request
request_data = {'model_id': fully_qualified_id}

• Replace the existing logic that prepends the currently active provider to the new model's ID with the updated code.

Verification

To verify the fix, test the model switcher dropdown by selecting a different provider and model, and check that the request sent to the gateway contains the correct fully-qualified model ID.

Extra Tips

• Ensure that the models list is correctly populated with the fully-qualified IDs for each model.
• Consider adding logging or debugging statements to verify that the correct ID is being extracted and used in the request.