Description
Claude Code is great, but confidential contracts or medical records cannot go to the cloud. Claude Code Local runs a small Python server that serves open models on your Mac's chip with MLX and answers Claude Code through Anthropic's own API, so nothing changes in use and nothing leaves the machine.
It ships several models including Gemma 4 31B, Qwen 3.5 122B at about 65 tok/s and DeepSeek V4 Flash with 1M context, running offline or air-gapped for legal and healthcare work.
Fully local:No cloud or API key.
Native API:Anthropic-format responses.
Model choice:Gemma, Qwen, DeepSeek and more.
Air-gapped:Works offline.
It ships several models including Gemma 4 31B, Qwen 3.5 122B at about 65 tok/s and DeepSeek V4 Flash with 1M context, running offline or air-gapped for legal and healthcare work.
Features
Fully local:No cloud or API key.
Native API:Anthropic-format responses.
Model choice:Gemma, Qwen, DeepSeek and more.
Air-gapped:Works offline.
