How to Integrate Hugging Face Models in Openclaw
Access thousands of open-source models through Hugging Face's Inference API. This guide walks you through authenticating and configuring Openclaw to use models like DeepSeek-R1.
Quick Start
Hugging Face provides free and paid access to state-of-the-art open models. With Openclaw, you can route requests through their inference endpoints without managing infrastructure.
Get Your API Token
First, create a fine-grained token in your Hugging Face account:
- Go to Settings → Tokens
- Create a fine-grained token
- Enable the Make calls to Inference Providers permission
Save this token as HF_TOKEN or HUGGINGFACE_HUB_TOKEN.
Configuration Options
Option A: Interactive Setup
Run the onboarding wizard:
Openclaw onboard --auth-choice huggingface-api-key
Option B: Non-Interactive Setup
For CI/CD or automated deployments:
Openclaw onboard --non-interactive \
--mode local \
--auth-choice huggingface-api-key \
--huggingface-api-key "$HF_TOKEN"
Using DeepSeek-R1
Once authenticated, configure your default model in ~/.Openclaw/Openclaw.json:
{
"agents": {
"defaults": {
"model": { "primary": "huggingface/deepseek-ai/DeepSeek-R1" }
}
}
}
Other popular models include meta-llama/Llama-3.3-70B and Qwen/Qwen2.5-72B.
Example Conversation
User: Explain quantum computing in simple terms
Assistant: Quantum computing uses qubits instead of classical bits. While a bit is either 0 or 1, a qubit can exist in a superposition of both states simultaneously. This allows quantum computers to process certain calculations exponentially faster than traditional computers.
Best Practices
- Use fine-grained tokens with minimal permissions for security
- Monitor your inference usage in the Hugging Face dashboard
- For production, consider the Inference Endpoints paid tier for better latency
Related: Check out our guide on Ollama Models Openclaw for running models locally, or explore vLLM Models Openclaw for high-throughput inference.
