Skip to content
Clean Coder Docs

Run Clean Coder on local model

Want privacy? Local LLMs are the answer.

Getting started

You can run Clean Coder with local LLMs in two ways:

Using Ollama:

  1. In your .env file, ensure there are no ANTHROPIC_API_KEY, OPENAI_API_KEY or OPENROUTER_API_KEY. If starting Clean Coder for the first time, skip the API key questions.
  2. Add OLLAMA_MODEL variable with your local LLM name. Example: OLLAMA_MODEL=quen2.5-coder:32b

Using local server (as LM Studio):

  1. Remove any external API keys from .env as mentioned above
  2. Add LOCAL_MODEL_API_BASE with your server URL. For LM Studio: LOCAL_MODEL_API_BASE=http://localhost:1234/api/v0
  3. Add LOCAL_MODEL_NAME with your model name. Example: LOCAL_MODEL_NAME=phi-4

Congratulations! Now you can enjoy autonomous code writing with full control and privacy.

🧠 Please keep in mind

Local models are still an experimental feature and might perform worse than top-tier commercial models.