Hugging Face Scraper - Models, Datasets, Papers

Scrape and export every AI model, dataset, space and daily research paper from the Hugging Face Hub — the world's largest open AI repository. This Hugging Face scraper hits the public HF Hub API, filters by task, library, language and license, and returns flat, export-ready rows with no token required.

🟢 No Hugging Face token. No login. No proxy. Export to JSON, CSV, Excel or XML.

---

What you get

Each record (model / dataset / space / paper) includes:

• entityType — model, dataset, space, paper or collection
• id / name — full repo ID and short name
• author — owning user or organization
• title / description — display title and summary
• downloads — download count (models / datasets)
• likes — community likes
• trendingScore — HF trending-algorithm score
• pipelineTag — primary task (text-generation, ASR, text-to-image, etc.)
• libraryName — transformers, diffusers, GGUF, MLX, ONNX, etc.
• tags / languages / datasets / baseModel — parsed from HF tag system
• license — extracted license (e.g. apache-2.0)
• gated / private / disabled — access flags
• createdAt / lastModified — timestamps
• fileCount — number of repo files
• readme — full model/dataset card (optional, fetchReadme)
• url — link to the Hub page
• For papers: paperId, paperAuthors, paperSummary, paperPublishedAt, paperUpvotes

---

Use cases

• AI / ML model tracking — monitor new and trending models for a task or library (e.g. all text-generation GGUF models) over time.
• Competitive AI intelligence — watch a specific org's releases (mistralai, meta-llama, stabilityai) by downloads and likes.
• Dataset discovery — enumerate training/evaluation datasets by language, task or license for sourcing.
• RAG / fine-tuning corpora — pull model and dataset cards (READMEs) at scale to index the Hub.
• Research & VC scouting — track daily papers and rising models to spot emerging AI trends early.

---

How to use

1. Click Try for free / Start.
2. Choose Entity Type — models, datasets, spaces, daily papers or collections.
3. Add filters: search, author/org, task (pipeline tag), library, language, tags, min downloads/likes.
4. Pick Sort (downloads, likes, recently updated, trending) and a Max Results cap.
5. Optionally enable Fetch Full Details / Fetch README for richer per-item data, then run and export as JSON, CSV, Excel or XML.

---

FAQ

Do I need a Hugging Face token or login?

No. The actor uses the public Hugging Face Hub API, which serves model, dataset, space and paper metadata anonymously. No token, account or proxy is required.

Can I scrape all models for a specific task or library?

Yes. Set the Task / Pipeline Tag (e.g. text-generation) and/or Library (e.g. gguf, diffusers), set Max Results to 0, and the actor auto-paginates the entire matching catalog.

Can I pull model cards / READMEs?

Yes. Enable Fetch Full Details and Fetch README to add the raw model/dataset card and richer fields (file list, license, gated status) to each record — ideal for RAG and indexing.

Which export formats are supported?

Each run is stored in an Apify Dataset and can be downloaded as JSON, JSONL, CSV, Excel, HTML or XML, or pulled via the Apify API.

Is this a Hugging Face API alternative?

Yes. It wraps the public Hugging Face Hub API and returns flat, export-ready rows, so it works as a no-login Hugging Face API alternative for teams that just want bulk metadata without writing API calls.

How do I export Hugging Face data to CSV or JSON?

Run the actor with your filters, then download the resulting dataset as CSV, JSON, JSONL, Excel or XML. This makes it a simple Hugging Face data export tool for models, datasets, spaces and papers.

Can I scrape Hugging Face without a token?

Yes. The actor reads the public Hub API anonymously, so you can scrape Hugging Face without a token, login or proxy. Use it as a Hugging Face dataset list scraper or daily papers scraper.

Changelog

2026-06-07

• Docs: added coverage for Hugging Face API alternative, exporting Hub data to CSV/JSON, and scraping Hugging Face without a token.