WikiArt API Alternative: Options for Paintings, Artists and Random Art
Developers look for a "WikiArt API" because WikiArt is one of the largest online collections of paintings, organised by artist, style, genre and period. WikiArt does have an API, but it comes with tight hourly limits, documentation from 2017 and no licence to reuse the artworks. This guide explains what WikiArt offers today and how to cover the same use cases with Art Search, the free museum APIs and Wikidata.
Art Search publishes this guide. Statements about WikiArt and the other services link to their own pages, checked on 22 November 2026.
Short answer
- WikiArt has a read-only API with a free tier of 400 requests per hour and paid packages billed by invoice. Its terms don't give you a licence to reuse the artworks it shows.
- For one API across many museums with natural-language search, filters, a random endpoint, commercial plans and SDKs, use Art Search.
- For images you may reuse commercially, use the CC0 works from museum open-access APIs such as the Met, the Art Institute of Chicago or Cleveland.
- For artist, style and movement metadata, use Wikidata.
Does WikiArt have an API?
Yes. WikiArt's API page describes a read-only API for information about artists and paintings, offered in four packages:
| Package | Price | Requests per second | Requests per hour | Sessions or IPs per hour |
|---|---|---|---|---|
| Free | Free (3, 6 or 12 month terms) | 4 | 400 | 10 |
| Academic | Free (3 month term) | 4 | 3,600 | 10 |
| Startup | $120 for 6 months, $200 for 12 months | 4 | 1,000 | 100 |
| Premium | $480 for 6 months, $840 for 12 months | 4 | 10,000 | Unlimited |
How access works
- Keys: you sign in to WikiArt to obtain an access code and a secret code. Paid packages are arranged by asking WikiArt for an invoice.
- Sessions: the documentation has your app call a login endpoint with both codes when it starts and add the returned session key (
authSessionKey) to later requests. - Documentation: a Google Doc linked from the API page. Its header marks it as obsolete, its version history has one entry (an initial version from January 2017), and it points to an even older document for operations it doesn't define. Its section on other search criteria is still a "TODO".
- Operations: painting search by term, paintings by artist, most viewed paintings, painting details, and lists of artists and "dictionaries" (WikiArt's grouping entries) with changes since a date.
- Paging: 60 items per page with a continuation token; you can't jump to a page number.
- Images: the
imageFormatparameter picks a thumbnail size, from 210 pixels wide up to 1920 by 1200.
Painting records include the title, artist, completion year, image URL and, in the detailed format, genres, styles, media, location and dimensions. That artist and style data is what most people want from WikiArt.
Terms, copyright and scraping
WikiArt's terms of use were last updated on 5 October 2016. They say WikiArt presents both public-domain and copyrighted artworks, and that it shows the copyrighted ones under the fair use principle: as historically significant works, for information and education, and as low-resolution copies "unsuitable for commercial use". WikiArt removes or blocks works when rights holders object.
For an app, that means:
- The terms don't grant API users a licence to the images. WikiArt's fair-use reasoning covers WikiArt's own site, not your product.
- Works by artists who died less than 70 years ago (the term in the EU and many other countries) are usually still in copyright. A commercial app needs permission from the rights holders, whatever API delivered the image.
- Scraping the website instead of using the API doesn't change any of this. WikiArt's
robots.txtdoesn't block crawlers and the terms don't mention automated access, but HTML pages aren't a stable interface, and the copyright question stays the same.
If your project is non-commercial, uses public-domain works and fits in 400 requests an hour, the WikiArt API can work. If not, the options below fit better.
Why developers want WikiArt data
Most requests for a WikiArt API come down to a few use cases:
- Browse by artist: all paintings by Monet, Hokusai or Rembrandt.
- Browse by style, movement or period: impressionism, ukiyo-e, the Dutch Golden Age, the 1890s.
- Images of famous paintings for an app, a quiz, a wallpaper or a slideshow.
- Random paintings for a daily artwork, a screensaver or a game.
- Search by subject or mood: "a stormy sea", "a quiet interior with a woman reading".
Art Search as a WikiArt alternative
Art Search is a REST API over more than one million artworks from museums worldwide: paintings, drawings, prints, sculptures, photographs and more. You search in natural language and filter by creation date, artwork type, material, technique, origin and image aspect ratio. All responses use one schema, there is a random artwork endpoint, and SDKs exist for 21 languages. The free plan has 50 requests a day for non-commercial use with a backlink; paid plans from $9 a month include commercial use (pricing).
Here is how the WikiArt use cases map to Search Artworks parameters. Every request costs one request from your daily quota.
| Use case | Art Search request |
|---|---|
| Style or movement | /artworks?query=impressionist river landscape&type=painting |
| Period | /artworks?query=landscape&type=painting&earliest-start-date=1860&latest-start-date=1890 |
| Subject or mood | /artworks?query=a quiet interior with a woman reading |
| Country or region | /artworks?query=woodblock print of a wave&origin=Japan |
| Technique or material | /artworks?query=harbour&type=print&technique=etching or material=oil_paint |
| Images that fit a layout | /artworks?query=mountain landscape&min-ratio=1.5 (landscape format) |
| A random artwork | /artworks/random |
| Details of one artwork | /artworks/{id} (title, image, dates, description) |
Artists: Art Search has no artist parameter and its responses have no artist field. You can name an artist in query, and the semantic search ranks related works first, but that is a ranking, not a filter: the results can include works by followers or works with a similar look. If you need a strict list of an artist's works, use a museum API with an artist filter or Wikidata.
Random paintings: the random endpoint returns any type of artwork, from paintings to swords. For a random painting, ask the search endpoint for one painting at a random position. This Python example shows both a filtered search and a random painting:
import os
import random
import requests
API_URL = "https://api.artsearch.io"
HEADERS = {"x-api-key": os.environ["ARTSEARCH_API_KEY"]}
def search(params):
response = requests.get(f"{API_URL}/artworks", params=params, headers=HEADERS, timeout=30)
response.raise_for_status()
return response.json()
# Style, subject and period in one request: impressionist river scenes from 1865 to 1900.
page = search({
"query": "impressionist river landscape",
"type": "painting",
"earliest-start-date": 1865,
"latest-start-date": 1900,
"number": 10,
})
for artwork in page["artworks"]:
print(artwork["id"], artwork["title"], artwork["image"])
def random_painting(query=None):
"""A random painting: one result at a random position (offset 0 to 1000)."""
params = {"type": "painting", "number": 1, "offset": random.randint(0, 1000)}
if query:
params["query"] = query
page = search(params)
if not page["artworks"] and page["available"]:
params["offset"] %= page["available"] # fewer matches than the offset: wrap around
page = search(params)
return page["artworks"][0] if page["artworks"] else None
print(random_painting("seascape at night")) Run it with ARTSEARCH_API_KEY=your-key python artsearch_examples.py. The tutorials go further: semantic art search in Python covers pagination and error handling, and the painting of the day widget shows a daily painting on a website without exposing your key. To try random artworks without code, open the random artwork generator.
What Art Search doesn't do: it has no artist, style or genre fields, no per-image licence information and no bulk download, and its terms don't allow storing the data.
Free museum APIs as WikiArt alternatives
Several museums publish their collections through free APIs, mostly under CC0. They cover one collection each, and several have the artist filter that Art Search lacks:
- The Met Collection API: CC0 data on more than 470,000 artworks, no key, and an
artistOrCulturesearch flag. Search returns object IDs, so details need one request per object. - Art Institute of Chicago API: no key, 60 requests per minute, IIIF images, Elasticsearch queries on fields such as
artist_titleandstyle_titles, and a documented recipe for a random public-domain artwork. - Cleveland Museum of Art Open Access API: CC0 data and images, an
artistsfilter, and full-size image files. - Rijksmuseum Search API: no key, a
creatorfilter, IIIF images; strong on Dutch painting. - Harvard Art Museums API: key required, non-commercial use only, random sort and colour data.
The comparison of the best art and museum APIs covers these and Europeana and the Smithsonian in detail, with rate limits and licences.
Wikidata and Wikimedia Commons
For the artist-and-style browsing WikiArt is known for, Wikidata is the closest open match. It describes paintings with their creator, date, movement, genre and collection, and links many of them to an image on Wikimedia Commons. Wikidata's data is CC0 (licensing); each Commons file has its own licence, and many paintings there are in the public domain (reusing Commons content).
This query lists paintings by Claude Monet that have an image. Q3305213 is "painting", Q296 is Claude Monet, and the image URL accepts a width parameter for a thumbnail:
import requests
QUERY = """
SELECT ?painting ?paintingLabel ?inception ?image WHERE {
?painting wdt:P31 wd:Q3305213; # instance of: painting
wdt:P170 wd:Q296; # creator: Claude Monet
wdt:P18 ?image. # image on Wikimedia Commons
OPTIONAL { ?painting wdt:P571 ?inception. }
SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
LIMIT 20
"""
response = requests.get(
"https://query.wikidata.org/sparql",
params={"query": QUERY, "format": "json"},
headers={"User-Agent": "MyArtApp/1.0 (https://example.com; you@example.com)"},
timeout=60,
)
response.raise_for_status()
for row in response.json()["results"]["bindings"]:
year = row.get("inception", {}).get("value", "")[:4]
print(row["paintingLabel"]["value"], year, row["image"]["value"] + "?width=400") The public SPARQL endpoint needs no key. Each query may run for 60 seconds, a client gets 60 seconds of processing time per minute, and requests without a descriptive User-Agent can be blocked (Query Service manual, User-Agent policy). Wikidata's coverage and image links vary from artist to artist, so check the results for the artists you need.
WikiArt API compared with the alternatives
| WikiArt API | Art Search | The Met | Art Institute of Chicago | Wikidata and Commons | |
|---|---|---|---|---|---|
| Access | Keys after sign-in, session login | API key | No key | No key | No key, User-Agent required |
| Free limit | 400 requests/hour | 50 requests/day | 80 requests/s | 60 requests/min | 60 s of query time per minute |
| Paid option | $200 or $840 per year, by invoice | $9 to $199 per month | None | None | None |
| Coverage | WikiArt.org | 1M+ artworks from museums worldwide | 470,000+ Met artworks | AIC collection | Paintings described by Wikidata editors |
| Browse by artist | Yes | Name in the query (ranking, not a filter) | Yes (artist or culture) | Yes (Elasticsearch fields) | Yes (SPARQL) |
| Style or movement | Yes | Describe it in the query | Keyword search | Yes (style fields) | Yes (movement property) |
| Natural-language search | No | Yes (semantic) | No | No | No |
| Random artwork | Not in the current documentation | Yes (endpoint) | No | Documented query recipe | Possible in SPARQL |
| Image licence | None granted; fair use on WikiArt | No per-image licence field | CC0 for public-domain works | CC0 for public-domain works | Per Commons file |
| Commercial use | Not covered for copyrighted works | API use on paid plans; image rights stay with the rights holders | Yes (CC0) | Yes (CC0 works) | Depends on each file's licence |
How to choose
- A non-commercial hobby project about painters: the WikiArt free tier or Wikidata can be enough. Stick to public-domain works.
- An app with search, a daily painting or random art, and a commercial plan: Art Search gives you one API across museums, natural-language search and a random endpoint.
- Print products, merchandise or anything that needs a clear image licence: CC0 works from the Met, the Art Institute of Chicago, Cleveland or the Smithsonian.
- Strict artist and movement lists or a research dataset: Wikidata, combined with museum data dumps.