🤗 Hugging Face Scraper - Models, Datasets & Spaces - PunkRecordsData

🚀 Export Hugging Face Hub data in seconds. Search AI models, datasets and Spaces and get structured rows with downloads, likes, trending score, task, license, gated status, parameter count and file lists. A single "llama" search returns models with 7.3M monthly downloads ranked for you. Export to CSV, Excel, JSON or XML.

The Hugging Face Scraper reads the Hub's own public API, so every number is exactly what the site shows: real download counts, likes, trending scores and per-model metadata across the largest open AI model repository. Filter by task (38 pipeline tags), author or keyword, sort by trending, downloads, likes or dates, and optionally pull full model details including license, parameter count and every file in the repo.

🎯 Target Audience	💡 Primary Use Cases
AI engineers and MLOps teams	Shortlist models by task with real download and license data
Analysts and VCs tracking AI	Measure model ecosystem momentum with trending scores
Compliance and legal teams	Audit licenses and gated status across the models you use
Researchers	Track datasets and Spaces around a research area

📋 What the Hugging Face Hub Scraper does

• Model records: id, author, task (pipeline tag), library, downloads, likes, trending score, gated status, tags, created and modified dates.
• Full model details (optional, per model): license, parameter count from safetensors, base model, file count and file list.
• Dataset records: the same search across Hub datasets, with downloads and likes.
• Space records: demo apps with SDK and likes.
• Filters that map 1:1 to the Hub: 38 task tags as a dropdown, author/organization, keyword search, five sort orders.

💡 Why it matters: model selection decisions need license, gating and adoption data side by side. The Hub shows them one page at a time; this actor gives you the whole comparison in one spreadsheet.

📊 Output of the Hugging Face model search

Real sample from a live run (22 fields per model with details on):

{
  "recordType": "model",
  "id": "meta-llama/Llama-3.1-8B-Instruct",
  "author": "meta-llama",
  "url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct",
  "likes": 8002,
  "downloads": 6079804,
  "trendingScore": 108,
  "gated": "manual",
  "pipelineTag": "text-generation",
  "libraryName": "transformers",
  "license": "llama3.1",
  "parameterCount": 8030261248,
  "baseModel": "meta-llama/Meta-Llama-3.1-8B",
  "fileCount": 17,
  "error": null
}

✨ Why choose this Hugging Face scraper

• 4 billable events, each switchable: models, datasets, Spaces and per-model details. No measured alternative offers more than one.
• License + parameter count + gated status in one row, the three fields every "can we use this model" decision needs.
• Real Hub numbers, read from the same API the site uses, not estimates.
• All three Hub surfaces in one actor instead of one actor per surface.
• Honest billing: details are charged only when the extra call actually returns; failures cost nothing.

📈 How this Hugging Face scraper compares to alternatives

Measured against the Hugging Face actors on the Apify Store (September 2026):

	This actor	Most-used alternative	Other HF actors
Billable data events	4 (models, datasets, spaces, details)	1	1
License / parameters / files per model	Yes, details event	No	No
Datasets and Spaces	Yes	Models only	Varies
Task filter	38-tag dropdown	Varies	Varies
Price per 1,000 models	$5.50	$5.00	$1.00 to $10.00

🚀 How to use the Hugging Face Hub Scraper

1. Create a free Apify account (with $5 of platform credit) at console.apify.com.
2. Open this actor's page and click Try for free.
3. Type search terms or an author (e.g. mistralai), pick a task if you want one.
4. Toggle datasets, Spaces or full model details as needed and press Start.
5. Download the dataset as CSV, Excel, JSON or XML.

💼 Business use cases

Model selection with legal sign-off

One run gives engineering the downloads and trending data and legal the license and gated status, for every candidate model at once.

AI market intelligence

Track which organizations ship models weekly and which tasks are heating up, using trending scores instead of anecdotes.

Dependency audits

List every file and the parameter count of the models your stack pulls, before they change under you.

Dataset discovery

Find datasets around your domain with adoption numbers, not just names.

🔌 Automating the Hugging Face Scraper

Connect to Make, Zapier, Slack, Airbyte, GitHub or Google Drive through Apify integrations: weekly trending sweeps to Slack, license-change monitoring, or warehouse syncs of your model watchlist.

🌟 Beyond business use cases

• Research: ecosystem studies of model releases by task and organization.
• Personal: watch your own models' downloads and likes over time.
• Non-profit: track open-license models suitable for public-interest work.
• Experimentation: a zero-setup Hub API playground.

🤖 Ask an AI assistant about this scraper

"I need a spreadsheet of Hugging Face models for a task, with downloads, license, gated status and parameter counts. Would the Hugging Face Scraper on Apify (apify.com/punkrecordsdata/huggingface-hub-scraper) do this on a schedule?"

❓ Frequently Asked Questions

🤖 How do I export a list of Hugging Face models to CSV or Excel?

Enter a search term or author, click Start, and download the dataset from the Storage tab in CSV, Excel, JSON or XML.

🔥 Can I get the trending models on Hugging Face?

Yes. Leave sort on "Trending" and the rows come back in the Hub's own trending order with the score included.

📜 How do I check the license of many models at once?

Enable "Full model details" and each model row carries its license, parameter count, base model and file list.

🔒 What does the gated field mean?

"No" means downloadable directly; "manual" or "auto" mirror the Hub's gated-access setting, where you must request access first.

📦 Does it cover datasets and Spaces too?

Yes, both are separate toggles and separate billable events, using the same search terms.

👤 Can I list everything from one organization?

Yes, set the author field (e.g. google) and leave search terms empty.

🧮 Are download numbers real?

They come from the Hub's own API, the same numbers the website renders.

💵 Do I pay for details on models where the call fails?

No. The details event is charged only when the extra call returns data.

📊 How many models can one run return?

Up to 1,000,000 on paid plans, following the Hub's own pagination. Free users get a 10-record preview.

⚙️ Does it need a Hugging Face API token?

No. It reads only public Hub data, no token required.

🗂 Can I filter models by task?

Yes, 38 official pipeline tags (text-generation, text-to-image, ASR and more) as a dropdown.

🔌 Integrate with any app

Datasets are available via the Apify API in JSON, CSV, Excel or XML for Python, Node.js, Sheets or BI tools, with webhooks on run completion.

🔗 Recommended Actors

• arXiv Research Papers Scraper - the papers behind the models, with citation metrics
• OpenCV Image Analyzer - deterministic computer vision on any image URL
• Steam Games Player Stats - another live-API data product
• Discord Scraper - community signals around AI projects

💡 Pro Tip: browse the complete PunkRecordsData collection for more data tools.

🆘 Need Help? contact.punkrecordsdata@gmail.com

⚠️ Disclaimer: independent tool, not affiliated with Hugging Face; only publicly available data.