# Hugging Face Scraper — Models, Datasets & Spaces with Stats (`highbrow_fame/hugging-face-hub`) Actor

Hugging Face Hub models, datasets and Spaces: downloads, likes, trending score, task, library, license, tags, gated flag, dates and a card summary. By search, author, task or repo link.

- **URL**: https://apify.com/highbrow_fame/hugging-face-hub.md
- **Developed by:** [yestrue](https://apify.com/highbrow_fame) (community)
- **Categories:** AI, Developer tools
- **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, 0 bookmarks
- **User rating**: No ratings yet

## Pricing

$0.80 / 1,000 repo delivereds

This Actor is paid per event. You are not charged for the Apify platform usage, but only a fixed price for specific events.

Learn more: https://docs.apify.com/actors/running/actors-in-store.md#pay-per-event

## What's an Apify Actor?

An Actor is a serverless cloud program that runs on the Apify platform. It has two run modes.
In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours,
and optionally produces a well-defined JSON output, datasets with results, or files in key-value store.
In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server.

Apify vocabulary and the platform model are defined once, in the agent quickstart at https://apify.com/agents.md.

## How to integrate an Actor?

If asked about integration, you help developers integrate Actors into their projects.
You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready.

Do not guess an integration path. Every one of them is in the agent quickstart at https://apify.com/agents.md: the Apify MCP server, Agent Skills with the Apify CLI, the JavaScript and Python clients, the REST API, and the account-free path for an agent with no human to sign in. It also carries the rule on stating cost before the first paid run.

For examples already wired to this Actor's own input schema, see the [API](#api) section below.

Each client library has reference documentation the quickstart does not restate: [JavaScript/TypeScript](https://docs.apify.com/api/client/js/docs.md) (`npm install apify-client`) and [Python](https://docs.apify.com/api/client/python/docs.md) (`pip install apify-client`).

# README

## Hugging Face Scraper — Models, Datasets & Spaces with Stats

Get models, datasets and Spaces from the Hugging Face Hub with their numbers: downloads in the last 30 days and all-time, likes, trending score, task (pipeline tag), library, license, tags, languages, base models, papers, gated flag, created and last-updated dates — and the first paragraph of each repo's card. Paste your huggingface.co search links, type search terms, authors, a task or a library — or list single repos.

**Why this one**

- 🔗 **Your search, as you set it up.** Set every filter you like on huggingface.co/models, /datasets or /spaces — task, library, language, license, parameter size, sort — and paste the link. Or type search terms and authors here and choose models, datasets or Spaces, a task, a library, tags and the sort order.
- ⚡ **Steady, within Hugging Face's limits, and no page limit.** A live test run read four searches of 250 (diffusers text-to-image models, MIT-licensed text-classification datasets, "flux" Spaces, Qwen's models) and two single repos — 1,002 repos, each with its card summary, in 114 seconds, without meeting the Hub's rate limit once. Without card summaries a run read 5,000 BERT models, most downloaded first, in under 3 minutes.
- 📝 **What each repo is, in one paragraph.** `cardSummary` is the first real paragraph of the repo's card, without the badges, link bars, notes and code around it. In the live run, 87% of the models and 88% of the datasets had one; most Spaces have no card text, and their short description stands in (12%).
- 🧭 **Forgiving input.** An author typed "qwen" found Qwen's models (the Hub itself matches authors case-sensitively). A profile link (huggingface.co/TheBloke) gives all that account's models; a repo link works with any sub-page (…/tree/main, …/discussions).
- 🔒 **No personal data beyond the public repo id.** The owner's handle is part of every repo id and is kept; nothing else about a person is read. In card summaries, the names of persons after "created by", "maintained by", "Authors:" and the like are replaced with `[name removed]` (the repo's own owner and organizations stay), @handles with `[handle removed]`, and e-mail addresses and phone numbers with `[e-mail removed]` / `[phone removed]`.
- 💸 **You pay only for repos delivered.** A link that is not a Hugging Face link, a repo that does not exist or an unknown author comes back as a free record with the reason.

### What you get

For every repo:

| Field | What it is |
|---|---|
| `id`, `type`, `author`, `name`, `url` | the repo (model, dataset or space) and its owner's handle |
| `pipelineTag`, `library` | the model's task (text-generation, text-to-image…) and library (transformers, diffusers, gguf…) |
| `downloads`, `downloadsAllTime` | downloads in the last 30 days and in total (models and datasets) |
| `likes`, `trendingScore` | as the Hub counts them at the time of the run |
| `createdAt`, `lastModified` | when the repo was created and last updated |
| `license`, `gated`, `gatedApproval`, `private`, `disabled` | the license, and whether access has to be requested (`auto` or `manual` approval) |
| `languages`, `baseModels`, `datasets`, `papers` | from the card's header and tags: languages, the models it was built on, training datasets, arXiv ids |
| `taskCategories`, `sizeCategory`, `modalities`, `formats` | for datasets |
| `title`, `sdk`, `runtimeStage`, `hardware`, `shortDescription` | for Spaces: gradio / docker / static…, RUNNING or SLEEPING, the hardware it runs on |
| `tags` | every tag the Hub shows |
| `cardSummary` | the first paragraph of the card (see above) |
| `input`, `position`, `ignoredFilters`, `scrapedAt`, `status`, `error` | which input it came from, its place in the results, and why an input gave nothing |

#### Example

A real record from a live run, 26 September 2026 (tags shortened):

```json
{
  "input": "meta-llama/Llama-3.1-8B-Instruct",
  "id": "meta-llama/Llama-3.1-8B-Instruct",
  "type": "model",
  "author": "meta-llama",
  "name": "Llama-3.1-8B-Instruct",
  "url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct",
  "pipelineTag": "text-generation",
  "library": "transformers",
  "downloads": 6184751,
  "downloadsAllTime": 186289232,
  "likes": 7888,
  "trendingScore": 85,
  "createdAt": "2024-07-18T08:56:00.000Z",
  "lastModified": "2024-09-25T17:00:57.000Z",
  "license": "llama3.1",
  "gated": true,
  "gatedApproval": "manual",
  "private": false,
  "languages": ["en", "de", "fr", "it", "pt", "hi", "es", "th"],
  "baseModels": ["meta-llama/Meta-Llama-3.1-8B"],
  "papers": ["arxiv:2204.05149"],
  "tags": ["transformers", "safetensors", "llama", "text-generation", "facebook", "meta", "…"],
  "cardSummary": "The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out). The Llama 3.1 instruction tuned text only models (8B, 70B, 405B) are optimized for multilingual dialogue use cases and outperform many of the available open source and closed chat models on common industry benchmarks.",
  "status": "OK",
  "scrapedAt": "2026-09-26T02:17:34.191Z"
}
```

### How to use it

1. Paste **Search links** from huggingface.co/models, /datasets or /spaces — or type **Search terms** and/or **Authors / organizations**, choose **Models, datasets or Spaces**, and optionally a **Task**, a **Library**, **Tags** and the **Sort** order. A task, library or tag alone is a search too (for example all `text-to-speech` models by downloads).
2. For single repos, list them under **Repos** — links or ids, one record each.
3. Set **Max results per search**, and untick **Add the card summary** if you only need the numbers (faster).
4. Press **Start**, and download the results from the **Output** tab as JSON, CSV or Excel.

To follow what is new or rising, save a search sorted by **Trending** or **Newest** as a task and schedule it daily.

You can also call it from the Apify API, from Make, Zapier or n8n, or from an AI agent through the Apify MCP server.

### Pricing

You are charged **per repo delivered** — see the price on this page. Card summaries cost nothing extra, only time. Links that are not Hugging Face links, repos that do not exist and unknown authors are free.

### FAQ

**How many results per search?** As many as the Hub has — it pages by cursor, 1,000 at a time, with no page limit; a live run read 5,000 BERT models in one search. The Hub does not say in advance how many there are.

**Which filters of a search link are used?** Search, author, task, library, language, license, other tags, dataset filters (task category, modality, size, format), parameter size, inference provider and app, and the sort. A few website-only filters (such as `base_model_relation` or `hardware`) have no counterpart in the Hub's API; they are left out and listed in `ignoredFilters` on each record of that search. A live test link for German conversational models up to 3B parameters, sorted by likes, came back with every result tagged `conversational` and `de`, in likes order, and none of the sampled models above 3B parameters.

**Why do Spaces have no downloads?** The Hub does not count downloads for Spaces; sorting Spaces by downloads sorts them by likes instead.

**Do I need a Hugging Face token?** No. Without one, the Hub allows one IP 500 API calls and 3,000 file reads per 5 minutes (its own rate-limit headers say so). The Actor reads those headers on every answer and spreads its requests over what is left of the window, so a normal run stays under the limit: a list call brings up to 1,000 repos, and each card summary is one file read — so over a long run, card summaries come at most about 3,000 per 5 minutes (untick **Add the card summary** for the numbers alone, which need only one call per 1,000 repos). If the Hub still asks it to slow down, the Actor waits as long as the Hub says and asks again; it never switches IPs to get around the limit (the proxy is only for network failures and server errors). With a token, the Hub applies your account's own, higher limits and the Actor paces by those; the token is sent only to huggingface.co, and also gives access to your private repos.

**Why is `cardSummary` empty for some repos?** Their card has no paragraph of text — only a header, a table, a list or links — or there is no card at all.

**Is this affiliated with Hugging Face?** No. This is an independent tool that reads public data from the Hugging Face Hub.

**Like it?** A short review on the Store page helps other people find this Actor. Something missing or broken? Tell us on the Issues tab — we read every one.

# Actor input Schema

## `searchUrls` (type: `array`):

Set up a search on huggingface.co/models, /datasets or /spaces with the filters you want (task, library, language, license, sort…) and paste the address. One per line.

## `repos` (type: `array`):

Or single repos, one record each: a link (https://huggingface.co/meta-llama/Llama-3.1-8B, …/datasets/openai/gsm8k, …/spaces/…) or an id (meta-llama/Llama-3.1-8B, datasets/openai/gsm8k). A profile link (https://huggingface.co/TheBloke) gives all of that account's models.

## `searchTerms` (type: `array`):

Or search here without a link: words in the repo name (llama, whisper, bert). Each is one search; the type and filters below apply.

## `maxItemsPerSearch` (type: `integer`):

Stops each search after this many repos.

## `authors` (type: `array`):

Accounts whose repos you want (meta-llama, Qwen, HuggingFaceFW). Each is one search; with search terms, every term is searched in every account.

## `repoType` (type: `string`):

For the search terms, authors and filters here.

## `task` (type: `string`):

Optional, as Hugging Face names it: text-generation, text-to-image, automatic-speech-recognition, image-text-to-text, feature-extraction, text-classification… For models the pipeline tag, for datasets the task category; Spaces have none.

## `library` (type: `string`):

Optional, for models and datasets: transformers, gguf, diffusers, sentence-transformers, mlx, peft, onnx… (datasets: datasets, pandas, polars…).

## `tags` (type: `array`):

Optional: any Hub tags, all of which must match — license:mit, en, conversational, modality:image, language:hu, size_categories:1K\<n<10K.

## `sort` (type: `string`):

For the search terms, authors and filters here (a search link keeps its own sort). Highest or newest first. Spaces have no downloads: they are sorted by likes instead.

## `includeCardSummary` (type: `boolean`):

Reads every repo's card (README) for its first paragraph. One more file per repo, so a little slower.

## `hfToken` (type: `string`):

Optional: a read token from huggingface.co/settings/tokens, for the Hub's higher limits for signed-in calls and for your own private repos. It is sent only to huggingface.co and never written to the log or the results.

## `maxConcurrency` (type: `integer`):

How many searches to work on at the same time.

## `proxyConfiguration` (type: `object`):

Used only when a request fails for network reasons or with a server error: the first try goes out directly. It is never used to get around Hugging Face's rate limits — those are waited for.

## Actor input object example

```json
{
  "searchUrls": [
    "https://huggingface.co/models?pipeline_tag=text-generation&library=gguf&sort=trending"
  ],
  "maxItemsPerSearch": 30,
  "repoType": "model",
  "sort": "trending",
  "includeCardSummary": true,
  "maxConcurrency": 3,
  "proxyConfiguration": {
    "useApifyProxy": true
  }
}
```

# Actor output Schema

## `results` (type: `string`):

Every result as a JSON record, with a status for each input.

## `resultsCsv` (type: `string`):

The same records as CSV, for a spreadsheet.

# API

You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup.

## JavaScript example

```javascript
import { ApifyClient } from 'apify-client';

// Initialize the ApifyClient with your Apify API token
// Replace the '<YOUR_API_TOKEN>' with your token
const client = new ApifyClient({
    token: '<YOUR_API_TOKEN>',
});

// Prepare Actor input
const input = {
    "searchUrls": [
        "https://huggingface.co/models?pipeline_tag=text-generation&library=gguf&sort=trending"
    ],
    "maxItemsPerSearch": 30,
    "proxyConfiguration": {
        "useApifyProxy": true
    }
};

// Run the Actor and wait for it to finish
const run = await client.actor("highbrow_fame/hugging-face-hub").call(input);

// Fetch and print Actor results from the run's dataset (if any)
console.log('Results from dataset');
console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`);
const { items } = await client.dataset(run.defaultDatasetId).listItems();
items.forEach((item) => {
    console.dir(item);
});

// 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs

```

## Python example

```python
from apify_client import ApifyClient

# Initialize the ApifyClient with your Apify API token
# Replace '<YOUR_API_TOKEN>' with your token.
client = ApifyClient("<YOUR_API_TOKEN>")

# Prepare the Actor input
run_input = {
    "searchUrls": ["https://huggingface.co/models?pipeline_tag=text-generation&library=gguf&sort=trending"],
    "maxItemsPerSearch": 30,
    "proxyConfiguration": { "useApifyProxy": True },
}

# Run the Actor and wait for it to finish
run = client.actor("highbrow_fame/hugging-face-hub").call(run_input=run_input)

# Fetch and print Actor results from the run's dataset (if there are any)
print(f"💾 Check your data here: https://console.apify.com/storage/datasets/{run.default_dataset_id}")
for item in client.dataset(run.default_dataset_id).iterate_items():
    print(item)

# 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/python/docs/quick-start

```

## CLI example

```bash
echo '{
  "searchUrls": [
    "https://huggingface.co/models?pipeline_tag=text-generation&library=gguf&sort=trending"
  ],
  "maxItemsPerSearch": 30,
  "proxyConfiguration": {
    "useApifyProxy": true
  }
}' |
apify call highbrow_fame/hugging-face-hub --silent --output-dataset

```

## MCP server setup

```json
{
    "mcpServers": {
        "apify": {
            "type": "http",
            "url": "https://mcp.apify.com/?tools=fetch-actor-details,highbrow_fame/hugging-face-hub"
        }
    }
}
```

The hosted server signs you in with OAuth on first connect, so no API token belongs in this config. Clients without OAuth support can send an `Authorization: Bearer <APIFY_API_TOKEN>` header instead, using a token from API & Integrations in Apify Console (https://console.apify.com/settings/integrations).

## OpenAPI specification

Download the OpenAPI definition: https://api.apify.com/v2/actors/OaYy3L1GNwhGPN5Vm/builds/xWJezFANllJBjGYaT/openapi.json