Hugging Face Models & Datasets Scraper

Exports Hugging Face model and dataset metadata — downloads, likes, tags, license, pipeline task, library, and update dates — by calling the official public huggingface.co API. Built for AI/ML engineers tracking model popularity or licensing, researchers building datasets of the Hugging Face ecosystem, and teams doing competitive or trend analysis on model releases.

Why this scraper

• Covers both models and datasets from a single actor run against the same official API — no need to run separate tools for each.
• Priced at $0.0003 per item, which is below the cheapest per-item alternative on the store.
• Pulls from the official unauthenticated JSON API, so output fields match what Hugging Face itself publishes (downloads, likes, trending score, license, tags) with no scraping of rendered HTML.

Output fields

Field	Type	Description
id	string	Full model or dataset ID, e.g. author/name
itemType	string	Either 'model' or 'dataset'
author	string	Organization or user that owns the item
pipelineTag	string	Pipeline task, e.g. text-generation (models only)
libraryName	string	Library used, e.g. transformers
downloads	integer	Total download count
likes	integer	Number of likes
trendingScore	integer	Hugging Face trending score
license	string	License identifier parsed from tags, e.g. mit
tags	string	Comma-separated list of all tags
gated	boolean	Whether access requires approval
private	boolean	Whether the item is private
sha	string	Latest commit SHA
createdAt	string	Creation timestamp, ISO 8601
lastModified	string	Last modification timestamp, ISO 8601
siblingsCount	integer	Number of files in the repo

Input

{
  "startUrls": [
    { "url": "https://huggingface.co/api/models?limit=50&full=true" }
  ],
  "maxItems": 50
}

To scrape datasets instead, point startUrls at a datasets endpoint, e.g. https://huggingface.co/api/datasets?limit=50.

Output

{
  "id": "Qwen/Qwen3.8-Flash-Next",
  "itemType": "model",
  "author": "Qwen",
  "pipelineTag": "image-text-to-text",
  "libraryName": "transformers",
  "downloads": 4810,
  "likes": 4176,
  "trendingScore": 4045,
  "license": "other",
  "tags": "transformers,safetensors,qwen4_exp,image-text-to-text,conversational,license:other,eval-results,endpoints_compatible,region:us",
  "gated": false,
  "private": false,
  "sha": "de4b8e4d43b917e7706784d8bb445c9af86a3540",
  "createdAt": "2026-08-24T08:24:59.000Z",
  "lastModified": "2026-08-27T05:03:36.000Z",
  "siblingsCount": 144
}

Pricing

$0.0003 per model or dataset exported. A run pulling 50 items costs $0.015; a run of 1,000 items costs $0.30.

Use cases

• Tracking weekly download and like counts for a shortlist of models to decide which to fine-tune or deploy.
• Building a license-compliance audit of datasets used internally by filtering exported records on the license field.
• Feeding a research database of pipeline tasks and library usage trends across newly released models.