Free The LLM
FTLLM is an Anthropic-compatible proxy that lets you use DeepSeek, Qwen, Llama, Gemini and hundreds more models in your favorite dev tools. Bring your own OpenRouter or NVIDIA NIM API key — no vendor lock-in, no markup.
Three steps to swap your coding AI for any open model.
Create an account in seconds
Bring your OpenRouter or NVIDIA NIM API key, pick a model
Set two environment variables and you're ready
Pick your tool and set two values — you are ready in under a minute.
# Run in your terminal:
export ANTHROPIC_AUTH_TOKEN=your-api-key
export ANTHROPIC_BASE_URL=https://ftllm.dev
# Then start Claude Code as usual
claudeGet your API key from the dashboard after signing up
Access hundreds of models through OpenRouter and NVIDIA NIM. Here are some popular picks — the full list is available in your dashboard.
Plus hundreds more — any model available on OpenRouter or NVIDIA NIM works with FTLLM.
Switch between models and providers freely. Your code, your choice.
Use your own OpenRouter or NVIDIA NIM API key. Pay only what your provider charges.
Set two environment variables and Claude Code or Cursor works with any model instantly.
Real-time streaming responses, thinking/reasoning blocks — everything works seamlessly.
Monitor your token usage, request counts and chat history from the dashboard.
Run FTLLM on your own infrastructure for complete control and privacy.
Sign up in seconds and start using any model in Claude Code or Cursor.