Free Community Hosted Instance of Docling-Serve on Hugging Face Spaces

[Bibyutatsu]

Hi everyone! 👋

I'm a big fan of docling and docling-serve. To make it easier for the community to quickly test and experiment with document parsing without setting up local infrastructure, I've deployed a free, public Hugging Face Space running docling-serve:

🚀 Live Space: Bibyutatsu/HF-docling-serve
📖 Swagger Docs: API Docs
䷛ Github Repo Github Repo

It runs the official Docker image (ghcr.io/docling-project/docling-serve:latest) on HF's Free Docker Tier (2 vCPU, 16GB RAM) with a thin wrapper that adds:

• ✅ Root / redirect for HF Spaces health checks
• ✅ Per-IP rate limiting (2 req/min on heavy endpoints)
• ✅ Global concurrency control (max 3 simultaneous tasks)

Quick Example

import requests

response = requests.post(
    "https://Bibyutatsu-HF-docling-serve.hf.space/v1/convert/source",
    json={"sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]}
)
print(response.json()['document']['md_content'])

Fair-Use Limits

Limit	Value
Per-IP Rate Limit	2 requests / minute on /v1/convert/* and /v1/chunk/*
Global Concurrency	Max 3 simultaneous heavy processing tasks
Cold Start	~2–5 min after 48h inactivity (free tier sleep)

For heavier workloads, the async endpoints (/v1/convert/source/async) queue efficiently and avoid timeouts.

Purpose

This is meant to help people quickly evaluate docling's layout detection, table extraction, and OCR capabilities on their own documents — without standing up their own infrastructure. The source repo with the Dockerfile and wrapper is here: Bibyutatsu/HF-docling-serve.

Would love to know if this could be useful to link from the community docs or README! Thanks again for the incredible work on this project! 📄✨