claude code 👀

Model qwen3-code, qwen3-ranker, qqen3-embedding deployed on VLLM docker image already available, ollama is easy to start with but too slow. Vscodium+continue or roocode add-on . hardware large vram 24gb more even better

LM Studio into Zed. You're on your own for figuring out a good model, sorry.

I’ve been doing Claude Code with 4.5 sonnet or opus The other harnesses have not been better (opencode) in my experience

If you aren't absurdly into it: Glm47-air on llamacpp into opencode

GPT-oss-20B on lmstudio on a gpu and feeding into a linux laptop with opencode. Not the smartest ine out there but has massive context window and it is quite fast.

#asknostr