Running open-source models without managing your own infrastructure just got easier. Here’s what you need to know:
- This matters because Hugging Face Inference API gives you access to thousands of open models without the DevOps overhead of self-hosting.
- Many developers struggle with token permissions and figuring out which models actually work through the inference endpoints.
- You will learn how to create the right token, configure Hugging Face in OpenClaw, and start using models like DeepSeek-R1.
Hugging Face hosts a massive collection of open-source models accessible through their Inference API. Instead of downloading multi-gigabyte models and setting up GPU servers, you call them via API.
First, create a fine-grained token in your Hugging Face account. Go to Settings → Tokens and enable the Make calls to Inference Providers permission. This is your HUGGINGFACE_HUB_TOKEN or HF_TOKEN.
For automated setups, configure non-interactively:
openclaw onboard --non-interactive --mode local --auth-choice huggingface-api-key --huggingface-api-key "$HF_TOKEN"
Set your preferred model in openclaw.json:
{
"agents": {
"defaults": {
"model": { "primary": "huggingface/deepseek-ai/DeepSeek-R1" }
}
}
}
Restart your gateway and you are ready to query open-source models through Hugging Face’s infrastructure.
