geepers-chat

One-command multi-provider LLM chat UI. Install, run, chat.

Supports 18 providers out of the box via geepers-kernel: Anthropic, OpenAI, xAI, Mistral, Cohere, Gemini, Perplexity, Groq, HuggingFace, Ollama, and more.

Quick start

pip install geepers-chat
export ANTHROPIC_API_KEY=sk-ant-...
python -m geepers_chat

Browser opens with a full chat UI. Pick a model, start chatting.

How it works

The package runs a local Flask server that:

1. Detects which providers you have API keys for (from env vars)
2. Serves a chat interface (the @geepers/chat widget)
3. Proxies your messages to provider APIs and streams responses back

No API keys leave your machine unless you're using gateway mode.

Gateway mode

If you don't have provider keys, the package can proxy through a gateway:

export GATEWAY_API_KEY=sk_...
python -m geepers_chat --gateway

Default gateway: https://api.dr.eamer.dev/v1

Options

python -m geepers_chat [options]

  --port <n>       Port (default: 3456)
  --host <addr>    Bind address (default: 127.0.0.1)
  --gateway [url]  Use gateway mode
  --no-open        Don't open browser

Environment variables

Set any of these to enable direct provider access:

License

MIT. Luke Steuber