1 related articles
Tested Qwen Code with custom OpenAI-compatible APIs: three env vars to start, modelProviders config, the 41k-character first request, doubled auto-memory requests, and retry hang pitfalls.
