Bug: explore tool hardcodes model to gpt-5.4-mini, ignoring custom/DeepSeek API configuration #3954

Describe the bug

After updating to the latest version of the GitHub Copilot CLI (v1.0.65), agent decision/attempt to invoke the explore tool ignores the configured custom model (e.g., DeepSeek endpoint configuration) and instead attempts to pass gpt-5.4-mini to the API endpoint. This causes upstream API providers like DeepSeek to fail with a 400 Bad Request because the model parameter is invalid for their endpoint.

✗ Explore(gpt-5.4-mini) Explore tracker button color flow
  └ No response generated

✗ 400 {"error":{"message":"The supported API model names are deepseek-v4-pro or deepseek-v4-flash, but you passed gpt-5.4-mini.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}} 

Affected version

GitHub Copilot CLI 1.0.65

Steps to reproduce the behavior

1. Configure GitHub Copilot CLI to use an external custom endpoint / DeepSeek API.
2. Run any command using the autonomous explore tool feature.
3. See error.

Expected behavior

The explore tool should respect the local configuration or environment variables defining the target model (e.g., deepseek-v4-pro), rather than falling back to a hardcoded default model name string during the tool call payload execution.

Additional context

• OS: Windows 11 / WSL2 (Ubuntu 24.04 LTS)