Liberate Your OpenClaw: A Guide to Hugging Face Inference Providers and Local Setup

Are you struggling to keep your OpenClaw agents running due to the recent limitations imposed by Anthropic on Claude models in open agent platforms? Don’t worry, there are alternative solutions available that can help you liberate your OpenClaw and keep your agents running smoothly.

Hugging Face Inference Providers: The Quickest Path Back to a Capable Agent

Hugging Face Inference Providers is an open platform that routes to providers of open-source models. It’s the perfect choice if you want access to the best models or don’t have the necessary hardware to run them locally.

To get started, you’ll need to create a token on the Hugging Face website. Once you have your token, you can add it to your OpenClaw configuration using the following command:

openclaw onboard --auth-choice huggingface-api-key

Paste your Hugging Face token when prompted, and you’ll be asked to select a model. We recommend using GLM-5 due to its excellent Terminal Bench scores, but there are thousands of models to choose from.

You can update your Hugging Face model at any time by entering its repository ID in the OpenClaw config:

{

agents: {

defaults: {

model: {

primary: "huggingface/zai-org/GLM-5:fastest"

}

}

}
}

Note that HF PRO subscribers get $2 free credits each month, which applies to Inference Providers usage. Learn more about this benefit here.

Local Setup: Full Privacy, Zero API Costs, and Full Control

If you prefer to run your models locally, you can use Llama.cpp, a fully open-source library for low-resource inference. To get started, install Llama.cpp using the following command:

brew install llama.cpp

Once installed, start a local server with a built-in web UI using the following command:

llama-server -hf unsloth/Qwen3.5-35B-A3B-GGUF:UD-Q4_K_XL

Here, we’re using Qwen3.5-35B-A3B, which works great with 32GB of RAM. If you have different requirements, please check out the hardware compatibility for the model you’re interested in. There are thousands of models to choose from.

To load the GGUF in Llama.cpp, use an OpenClaw config like this:

openclaw onboard --non-interactive

--auth-choice custom-api-key

--custom-base-url "http://127.0.0.1:8080/v1"

--custom-model-id "unsloth-qwen3.5-35b-a3b-gguf"

--custom-api-key "llama.cpp"

--secret-input-mode plaintext

--custom-compatibility openai

Verify the server is running and the model is loaded using the following command:

curl http://127.0.0.1:8080/v1/models

Which Path Should You Choose?

Use Hugging Face Inference Providers if you want the quickest path back to a capable OpenClaw agent. Use Llama.cpp if you want privacy, full local control, and no API bill. Either way, you don’t need a closed hosted model to get OpenClaw back on its feet!