Hugging Face API

Hugging Face models, datasets, Spaces, papers and profiles as clean JSON, with filters for task, license and size.

The API.SH Hugging Face API returns public Hub metadata as JSON. Search models by task, library, license, language, dataset, base model and parameter count, and read a model's downloads, likes, trending score, tags, architectures, files and card metadata. Do the same for datasets and Spaces, browse collections and Daily Papers, read a paper with its abstract, AI summary and citing models, and look up a user or organization profile.

Get a free API key Try it in the playground

11 endpoints
from $1 / 1,000 calls
1,000 free credits / month
failed calls free

Hugging Face dataset search

GET https://api.api.sh/huggingface/v1/datasets

{ "success": true }

1 credit · $1 per 1,000 calls

What you can build with the Hugging Face API

Model discovery and comparison

Filter models by task, library, license, language and parameter count, sorted by downloads, likes or trending score.

/models /model

Dataset search

Find datasets by task, language, license, modality, format and size, and read card metadata, splits and row counts.

/datasets /dataset

Trending and market tracking

Track which models, datasets and Spaces are gaining likes and downloads, and which organizations publish them.

/models /spaces /profile

Research paper monitoring

Follow the Daily Papers list by day, week or month, search papers, and see which models, datasets and Spaces cite each one.

/papers /papers/search /paper

Demo and Space analysis

List ML demo apps by SDK, model or dataset and read each Space's runtime stage, hardware and linked models.

/spaces /space

Curated collections and AI agents

Read curated collections and their items, or give an agent Hub search so it can pick models for a task.

/collections /models

Hugging Face data you get

Field names from real responses. Numbers come parsed, URLs absolute, dates in ISO 8601.

Hugging Face API endpoints and prices

Every Hugging Face endpoint costs 1 credit per call: credits are $1 per 1,000 on a monthly plan, $2 pay-as-you-go. Pricing

Search and discovery4

Hugging Face dataset search /huggingface/v1/datasets	$1 · $2 PAYG 1 credit
Hugging Face model search /huggingface/v1/models	$1 · $2 PAYG 1 credit
Hugging Face paper search /huggingface/v1/papers/search	$1 · $2 PAYG 1 credit
Hugging Face Spaces search /huggingface/v1/spaces	$1 · $2 PAYG 1 credit

Profiles and accounts1

Hugging Face user or organization

/huggingface/v1/profile

$1 · $2 PAYG

1 credit

More data6

Hugging Face collections /huggingface/v1/collections	$1 · $2 PAYG 1 credit
Hugging Face dataset /huggingface/v1/dataset	$1 · $2 PAYG 1 credit
Hugging Face model /huggingface/v1/model	$1 · $2 PAYG 1 credit
Hugging Face paper /huggingface/v1/paper	$1 · $2 PAYG 1 credit
Hugging Face Daily Papers /huggingface/v1/papers	$1 · $2 PAYG 1 credit
Hugging Face Space /huggingface/v1/space	$1 · $2 PAYG 1 credit

Call the Hugging Face API

Hugging Face dataset search: GET https://api.api.sh/huggingface/v1/datasets

curl "https://api.api.sh/huggingface/v1/datasets?q=squad" \
  -H "x-api-key: $APISH_KEY"

import os, requests

res = requests.get(
    "https://api.api.sh/huggingface/v1/datasets",
    params={"q": "squad"},
    headers={"x-api-key": os.environ["APISH_KEY"]},
    timeout=120,
)
res.raise_for_status()
data = res.json()

const url = new URL("https://api.api.sh/huggingface/v1/datasets");
url.searchParams.set("q", "squad");

const res = await fetch(url, { headers: { "x-api-key": process.env.APISH_KEY } });
if (!res.ok) throw new Error(`${res.status} ${await res.text()}`);
const data = await res.json();

# Google Sheets (API.SH add-on)
=APISH("/huggingface/v1/datasets", "q", "squad")

# Zapier, Make, n8n (verified node): pick the API.SH app,
# choose "Hugging Face" and "Hugging Face dataset search".

# AI agents (Claude, ChatGPT, Cursor): add the
# API.SH MCP server https://api.sh/mcp

Create a free accountConfirm your email and get 1,000 free credits every month. No card needed.
Create an API keyOne key works for all 483 platforms.
Send the requestJSON back in one call. Add output=csv or output=ndjson for files.

Why API.SH for Hugging Face data

Public data only: no Hugging Face login, cookies or account needed
One key and one response shape for Hugging Face and 482 other platforms
Built and maintained by API.SH, not third-party publishers
Pay per call; errors on our side cost nothing
Live status for every endpoint

API.SH vs the official Hugging Face API

Hugging Face has official, documented Hub API endpoints and a separate Inference Providers API for running models with an access token. API.SH covers the first kind of data only: it returns Hub metadata as JSON with no token and does not run models.

Skip writing your own Hub scraper or paginating raw Hub calls and Apify actors: API.SH gives you one key and a fixed price per call.

Hugging Face API questions

What Hugging Face data can I get?+

Model search and detail (/huggingface/v1/models, /huggingface/v1/model), datasets (/huggingface/v1/datasets, /huggingface/v1/dataset), Spaces (/huggingface/v1/spaces, /huggingface/v1/space), Daily Papers and paper search (/huggingface/v1/papers, /huggingface/v1/papers/search, /huggingface/v1/paper), collections (/huggingface/v1/collections) and user or organization profiles (/huggingface/v1/profile).

How much does the Hugging Face API cost?+

Every Hugging Face endpoint costs 1 credit per call. On a monthly plan 1,000 credits cost $1 (Starter is $50/month for 50,000 credits). Pay-as-you-go is $2 per 1,000 credits, with top-ups from $15 that never expire.

Is there a free trial?+

Yes. Every account gets 1,000 free credits every month. That is 1,000 calls to /huggingface/v1/collections at 1 credit each.

Is there an official Hugging Face API?+

Yes. Hugging Face documents Hub API endpoints and a separate Inference Providers and Inference Endpoints service for running models, which use access tokens. API.SH only returns public Hub metadata such as models, datasets, Spaces and papers, and does not run models.

Do I need a Hugging Face account or access token?+

No. API.SH reads public Hugging Face pages and Hub data only. No endpoint uses a login, an access token or anyone's account.

How do I get started?+

1. Sign up at api.sh. 2. Copy your API key. 3. Call /huggingface/v1/models with q set to a model name or task. Request examples and every parameter are in the docs at https://api.sh/docs. You can also use the JS or Python SDK, Google Sheets, Zapier, Make, n8n or the MCP server at https://api.sh/mcp.

Is there a free Hugging Face API, and is this the Inference API?+

This is not the Inference API: it does not run models or generate text or images. It returns Hub metadata about models, datasets, Spaces and papers. Every account gets 1,000 free credits every month. That is 1,000 calls to /huggingface/v1/collections at 1 credit each.

Can I filter Hugging Face models by license, size or task?+

Yes. /huggingface/v1/models filters by task, library, language, license, dataset, base model, minimum and maximum parameter count and tags, with sorting and cursor paging. /huggingface/v1/datasets filters by task, language, license, modality, format and size.