I've been trying to do LLM coding with Continue in VS Code without much success. It felt like the models I was trying and Continue just didn't talk the same language. I found the problem was that continue seemed to support a limited set of local models. Last night I switched to OpenCode, and was able to start getting work done. I'll try to get to the point were I can automate the comparison of work done by different models, and hopefully present some useful info. 🙂