claude-code-local
Claude Code Local is a MLX server based on Python that allows Claude Code to be run entirely locally on Mac computers with Apple Silicon processor. Without the need for a cloud connection or a API key, Anthropic matches the API format with local model outputs and converts vehicle call formats. The system allows you to perform code editing and terminal operations by offering intersessional prompt caching support.
nicedreamzapp · Local AI / Self-Hosted LLM
Common commercial products in this category: ChatGPT, Claude, Gemini
- On your Apple Silicon Mac computer, you run Claude Code with local AI models without a cloud connection or a API key.
- Thanks to the Python-based MLX server, you make the vehicle call formats of local models compatible with Claude Code.
- With automatic installation scripts and system control tools, you select and install the model suitable for Mac memory capacity.
- ↑ PositiveGitHub Issue
Qwen3.6-35B-A3B Support
“### What would you like to see? Hello, would this model support Qwen3.6-35B-A3B? It’s very powerful and might be a better fit for people running 64gb or less of unified memory. ### Why is this useful? _No response_ ### Your Mac model + RAM (if relevant) M4 Studio Ultra 64GB Ram”
- • NeutralGitHub Issue
Gemma 4 12B support request
“### What would you like to see? Hello, Is there any possibility to add support for the new Gemma 4 12B for weaker laptops like m1 pro 32gb? Or maybe publish an abliterated version and consumers can use it via the existing environment variables. ### Why is this useful? Weaker…”