Skip to content

Instantly share code, notes, and snippets.

@kibotu
Created March 17, 2026 08:51
Show Gist options
  • Select an option

  • Save kibotu/2fa719cd83b6a4c0e424d4ed7128a262 to your computer and use it in GitHub Desktop.

Select an option

Save kibotu/2fa719cd83b6a4c0e424d4ed7128a262 to your computer and use it in GitHub Desktop.
Local Proxy to use claude on mac and connect to remote ollama server.
# Use the actual model name from `ollama list`
export LM_MODEL="qwen3.5:35b" # or qwen3-coder:30b for coding tasks
export ANTHROPIC_BASE_URL="http://127.0.0.1:8131" # point to local proxy
export ANTHROPIC_AUTH_TOKEN="local"
export CLAUDE_CODE_MAX_OUTPUT_TOKENS=128000
export ANTHROPIC_MODEL="$LM_MODEL"
export ANTHROPIC_DEFAULT_OPUS_MODEL="$LM_MODEL"
export ANTHROPIC_DEFAULT_SONNET_MODEL="$LM_MODEL"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="$LM_MODEL"
export CLAUDE_CODE_SUBAGENT_MODEL="$LM_MODEL"
2. Start the proxy (run once, or add to a startup script)
bashpip install 'litellm[proxy]'
litellm --model ollama/qwen3.5:35b \
--api_base http://1.1.1.1:11434 \
--host 127.0.0.1 \
--port 8131 \
--drop_params
Then in a new terminal just run claude normally.
Optional: auto-start the proxy in .zshrc
bash# Auto-start litellm proxy if not already running
if ! lsof -i :8131 -sTCP:LISTEN &>/dev/null; then
litellm --model ollama/qwen3.5:35b \
--api_base http://1.1.1.1:11434 \
--host 127.0.0.1 \
--port 8131 \
--drop_params \
> ~/.litellm.log 2>&1 &
echo "🦙 litellm proxy started (ollama → :8131)"
fi
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment