Hugging Face Hub Scraper — Models, Datasets & Spaces API

SEO title: Hugging Face Hub Scraper | Models, Datasets & Spaces Downloads, Likes & Licenses
SEO description: Scrape Hugging Face models, datasets, and Spaces via the official Hub API. Get downloads, likes, trending score, licenses, gated flags, tasks, files, and a health score — no API key.

One Actor for the whole Hub. Search or look up models, datasets, and Spaces from the public Hugging Face Hub REST API and export clean JSON/CSV for AI research, competitive monitoring, RAG source discovery, and agent tooling.

Why this beats typical Store competitors

Most Hub Actors cover models only (or split models/datasets into separate paid Actors) and stop at raw metadata. This one:

• Scrapes models + datasets + Spaces in a single run
• Adds a healthScore (popularity + freshness − gated penalty)
• Parses licenses, pipeline/task tags, languages, file counts, architectures, safetensors totals, Space SDK/title
• Supports exact repo ID enrichment plus Hub list pagination via Link cursors
• Uses the official public API — no browser, no proxies, no key required for public repos

Example input

{
  "resourceTypes": ["models", "datasets", "spaces"],
  "search": "llama",
  "sort": "downloads",
  "direction": "-1",
  "maxResultsPerType": 10,
  "includeFull": true,
  "includeConfig": true
}

Example output (truncated)

{
  "resourceType": "model",
  "id": "meta-llama/Llama-3.1-8B-Instruct",
  "author": "meta-llama",
  "downloads": 1234567,
  "likes": 8900,
  "trendingScore": 12,
  "pipelineTag": "text-generation",
  "libraryName": "transformers",
  "license": "llama3.1",
  "gated": "auto",
  "fileCount": 18,
  "hubUrl": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct",
  "healthScore": 84.2,
  "source": "huggingface.co"
}

Suggested pay-per-event pricing

Event	Suggested price
Actor start	$0.0005
Per Hub item (model / dataset / Space)	$0.002 – $0.005

At $0.003/item and 100k paid rows/month → ~$300 revenue → ~$240 after Apify’s 20%, minus tiny HTTP compute.

Data source & limits

• API docs: https://huggingface.co/docs/hub/en/api
• Public list endpoints paginate with Link: rel="next" cursors; this Actor follows them until maxResultsPerType.
• Subject to Hugging Face rate limits; for heavy production traffic consider an optional HF token later (not required for this Actor).
• Not affiliated with Hugging Face.

Changelog

• 1.0.0 — Unified models/datasets/Spaces scraper with health scores and Store-ready schema.

Use cases

• AI researchers & founders — track downloads/likes for models in a task or library; spot rising GGUF / multimodal repos.
• ML platform teams — inventory gated vs open models and licenses before adopting Hub deps.
• Competitive intel — watch Spaces and datasets for a search query on a schedule.
• Apify MCP / agents — field-level dataset metadata helps agents pick hubUrl, downloads, and healthScore.

Buyer task: ai-lab-hf-trending-models. Examples: https://github.com/LMH-Solutions/apify-open-data-actors (examples/hf_trend_watch.py)