Self-hosted HuggingFace mirror

Your models,
served locally.

This mirror caches models, datasets and spaces from HuggingFace behind your own network — client machines pull everything from here with no direct internet access needed.

Point any client at the mirror by setting HF_ENDPOINT — every HuggingFace library and CLI honors it.

terminal — set your mirror endpoint

$ export HF_ENDPOINT=http://localhost:8090

$ export HF_ENDPOINT=http://localhost:8090

$env:HF_ENDPOINT = "http://localhost:8090"

From now on, every download in the huggingface_hub library flows through this mirror.

Browse cached repositories →

Download anything

The huggingface-cli works exactly as it would against huggingface.co — just pointed here.

install the cli

$ pip install -U huggingface_hub

download a model

$ huggingface-cli download --resume-download openai-community/gpt2 --local-dir gpt2

download a dataset

$ huggingface-cli download --repo-type dataset --resume-download Salesforce/wikitext --local-dir wikitext

from python — snapshot_download()

from huggingface_hub import snapshot_download

snapshot_download(repo_id='Qwen/Qwen-7B', repo_type='model',
    local_dir='./model_dir', resume_download=True,
    max_workers=8)