Comment by seanmcdirmid

I actually considered lite LLM to get Qwen to work with codex. It mostly works without lite LLM but MCP servers aren’t available, lite LLM tricks into think Qwen is an OpenAI model.