GitHub radar
Run your own local code ai without paying a monthly fee!
A smaller version of Qwen’s code generation model that can run offline on your computer rather than using a paid cloud service.
It's a trimmed-down large language model for writing code: half of its internal building blocks were removed so it fits on a regular computer instead of needing a server.
When it helps
Useful if you want a free, local assistant for simple, repetitive coding tasks; for long, complex multi-step work (like Claude Code on Opus) it's noticeably weaker.
Pros
- Runs locally — no subscription and no internet connection needed.
- Fits in memory on one 32GB graphics card or Mac.
- Loses only about 1% quality on simple coding tasks.
- Still understands images, not just code.
Cons
- Quality drops about 9% on long, multi-step coding tasks.
- Needs a powerful GPU or Mac with 32GB+ memory.
- It's an experimental release; the authors still want feedback.
How to set it up — step by step
- 1Open your AI agent (Claude Code, Codex) inside your project folder.
- 2Send the agent the text from the block below — it installs via Ollama or llama.cpp.
- 3Check you have a 32GB graphics card or a Mac with 32GB+ memory.
- 4Ask the agent to show you how to give it coding tasks.
- 5Give the model a simple coding task and compare the result.
- 6For long, complex tasks, switch back to your regular large model.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up this coding model locally for me: https://huggingface.co/ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-Coder-GGUF - it's a ready GGUF file, run it through Ollama or llama.cpp (needs a Mac with 32 GB of memory or a 32 GB GPU), and show me how to give it coding tasks.
You need a Mac with at least 32 GB of unified memory, or a GPU with 32 GB of VRAM - that is how much space the compressed model takes up.
Open on Hugging Face▌ More finds