The Best Art and Museum APIs in 2026

Last updated · Art Search

If you need artwork data or images in an app, you can choose between free open-access APIs run by individual museums, free aggregators such as Europeana and the Smithsonian that cover many institutions, and commercial APIs that search across many collections at once. This guide compares the main options on what decides a project: coverage, API keys, rate limits, data and image licences, IIIF support and how search works.

Art Search publishes this guide and is one of the APIs compared. Statements about other providers link to their documentation or, where a number is not documented, were checked against the live API on 22 November 2026. Limits and terms change, so confirm the details in the linked docs before you build on them.

Quick answer: which art API should you use?

Art and museum APIs compared

The table summarises each API. The sections below explain every cell and link to the source.

Art and museum APIs at a glance (checked 22 November 2026)
API Coverage API key Published limits Data licence Images Search
Art Search 1M+ artworks from museums worldwide Yes (free plan) Free: 50/day, 1/s. Paid: 500 to 50,000/day plus overage Commercial use on paid plans; no per-item licence field One image URL per artwork; no IIIF Natural-language (semantic) query plus filters
The Met 470,000+ Met artworks No 80 requests/s CC0 JPEG for public-domain works Keyword plus filters; returns object IDs
Art Institute of Chicago AIC collection No 60 requests/min (anonymous) Artwork data CC0 (descriptions CC BY 4.0) IIIF Image API 2.0 Elasticsearch query DSL
Cleveland Museum of Art 64,000+ records, 37,000+ with images No Not published CC0 JPEG (web, print) and TIFF Keyword plus filters; image similarity
Rijksmuseum Rijksmuseum collection No Not published Mostly public domain or CC0; some CC BY 4.0 IIIF Field search; returns Linked Data IDs
Harvard Art Museums Harvard Art Museums collection Yes (request form) 2,500 calls/day; non-commercial only Terms of use (non-commercial) IIIF Image API 2.1 Keyword, Elasticsearch, colour, random sort
Europeana Records from European libraries, archives and museums Yes (personal key, instant) Not published; project keys for services Metadata CC0; media per item Thumbnails; IIIF where the institution provides it Keyword, rights, colour, random sort
Smithsonian All Smithsonian museums, not only art Yes (api.data.gov) api.data.gov default: 1,000/hour CC0 (Open Access) Thumbnail, screen and high-res JPEG Keyword, art and design category, random sort
WikiArt WikiArt.org paintings and artists Keys after sign-in Free: 400/hour; paid up to 10,000/hour No reuse licence; copyrighted works shown under fair use Several thumbnail sizes Term search, by artist, most viewed
Wikidata and Commons Artworks, artists, movements; media files No (User-Agent required) SPARQL: 60 s per query, 60 s per minute Wikidata CC0; Commons per file Commons files and thumbnails SPARQL

Art Search is a commercial REST API from skycraft GmbH in Dresden. It indexes more than one million artworks from museums worldwide, including paintings, sculptures, prints, drawings, photographs, furniture and textiles, and returns them in one schema. Its search is semantic: you describe what you want in plain language, such as "a stormy sea at night" or "a melancholy portrait in blue tones", and filters narrow the results by date, type, material, technique, origin and image aspect ratio.

Coverage
More than one million artworks; 24 artwork types such as painting, drawing, print, sculpture, photograph, tapestry and calligraphy.
Endpoints
Search Artworks (GET /artworks), Retrieve Artwork (GET /artworks/{id}) and Random Artwork (GET /artworks/random). Each call costs one request.
API key
Required, sent as the api-key query parameter or the x-api-key header. The free key comes from the console.
Limits
Free: 50 requests a day, 60 requests per minute, 1 concurrent request. Apprentice ($9/month): 500 a day, then $0.006 per request, 2 per second. Artisan ($39): 5,000 a day, then $0.004, 10 per second. Maestro ($199): 50,000 a day, then $0.002, 20 per second. See pricing.
Licence
Paid plans include commercial use. The free plan is for non-commercial use and requires a backlink. Responses have no licence or rights field per artwork, and the terms don't allow storing API data.
Images
One image URL per artwork. No IIIF.
Search
Natural-language query plus earliest-start-date, latest-start-date, earliest-end-date, latest-end-date, min-ratio, max-ratio, type, material, technique and origin. Up to 10 results per request (number), offset from 0 to 1,000.
Tooling
SDKs in 21 languages and an OpenAPI 3 specification.
Best for
Apps that search across collections by meaning, need one consistent schema, show random artworks, or are commercial products that need a supported API.

The example request from the documentation searches for medieval knights:

curl "https://api.artsearch.io/artworks?query=medieval+knights&number=3" \
  -H "x-api-key: YOUR-API-KEY"
{
    "available": 993,
    "number": 3,
    "offset": 0,
    "artworks": [
        {
            "id": 25623857,
            "title": "Knightly Sword",
            "image": "https://img.artsearch.io/artworks/knightly_sword_unknown23_1450.jpg"
        },
        {
            "id": 25426886,
            "title": "Sculpture of a Kneeling Knight or King",
            "image": "https://img.artsearch.io/artworks/sculpture_of_a_kneeling_knight_or_king_unknown383_1250.jpg"
        },
        {
            "id": 25934005,
            "title": "A Knight in Armor",
            "image": "https://img.artsearch.io/artworks/a_knight_in_armor_unknown32_1907.jpg"
        }
    ]
}

Retrieving an artwork by id adds start_date, end_date and a description. The random endpoint returns the same fields for one random artwork; the random artwork generator shows it in action.

Where Art Search is weaker: there is no filter by museum and no artist field, a query returns at most 10 results per request and 1,010 in total, there is no bulk download, and you may not store the data. If you need a full dataset or per-image rights, the museum APIs below are the better choice.

The Met Collection API (Met Museum API)

The Metropolitan Museum of Art publishes data on more than 470,000 artworks for unrestricted commercial and non-commercial use under Creative Commons Zero (CC0), with high-resolution JPEG images of the works that are in the public domain. You don't need to register or get a key; the Met asks you to stay under 80 requests per second (Met Collection API docs).

Coverage
More than 470,000 artworks in the Met's collection, across all curatorial departments.
API key
None.
Limits
The Met asks you to stay under 80 requests per second.
Licence
CC0 for the Open Access data. Images are included for public-domain works (isPublicDomain); use is also subject to the Met's terms and conditions.
Images
primaryImage, primaryImageSmall and additionalImages as JPEG URLs. The docs don't describe a IIIF service.
Search
Keyword q with filters isHighlight, title, tags, departmentId, isOnView, artistOrCulture, medium, hasImages, geoLocation and dateBegin/dateEnd. Search returns object IDs only, so each artwork's details need a second request.
Bulk data
The Open Access CSV on GitHub.
Best for
Projects about the Met's collection, CC0 data, prototypes without a key.

The old /public/collection/v1/search endpoint was retired on 1 October 2026 and now answers with HTTP 410. Its replacement, /public/collection/v1.1/search, is paginated: 100 IDs per page by default, up to 500 with limit, and only the first 10,000 results of a query are reachable.

# Object IDs of matching artworks with images
curl "https://collectionapi.metmuseum.org/public/collection/v1.1/search?q=sunflowers&hasImages=true&limit=20"

# One object: title, artist, dates, images, isPublicDomain and more
curl "https://collectionapi.metmuseum.org/public/collection/v1/objects/436524"

Art Institute of Chicago API

The Art Institute of Chicago's API is the same one that runs the museum's website and mobile app. It needs no key, returns JSON and supports Elasticsearch queries. Images come from a separate server that implements the IIIF Image API 2.0 (AIC API docs).

Coverage
The AIC collection: artworks plus agents, places, exhibitions, galleries and other resources.
API key
None. AIC asks you to send an AIC-User-Agent header with your project name and a contact email.
Limits
Anonymous users are throttled to 60 requests per minute. Listing and search return at most 100 records per page and 1,000 records per query.
Licence
Artwork data is CC0, except the description field, which is CC BY 4.0; other resources such as places and website content have their own terms (see the API docs). Images of public-domain works are CC0 (image licensing); AIC recommends using only images of artworks tagged is_public_domain.
Images
IIIF Image API 2.0 at https://www.artic.edu/iiif/2/{image_id}/full/843,/0/default.jpg. Larger widths (1,200 and 1,686 pixels) are available for public-domain images only. IIIF manifests exist for public-domain artworks.
Search
Full-text q plus the Elasticsearch Query DSL, fields to trim responses, and aggregations. The docs include a recipe for a random public-domain artwork using random_score.
Bulk data
Data dumps on GitHub, updated weekly.
Best for
AIC projects, IIIF viewers such as Mirador, CC0 data with rich search.
curl -g "https://api.artic.edu/api/v1/artworks/search?q=water%20lilies&query[term][is_public_domain]=true&fields=id,title,image_id&limit=5" \
  -H "AIC-User-Agent: my-art-app (you@example.com)"

Cleveland Museum of Art API (Open Access)

The Cleveland Museum of Art provides data on more than 64,000 artwork records under CC0 and image files for more than 37,000 works under the same terms. Each open-access work links to a web JPEG, a print-size JPEG and a full-size TIFF (CMA Open Access API docs).

Coverage
64,000+ artwork records; 37,000+ works with images.
API key
None mentioned in the docs; the examples call the API without credentials.
Limits
No rate limit is published; check the provider's docs.
Licence
All data is CC0. Images are CC0 for works whose share_license_status is CC0.
Images
Direct file URLs: web JPEG, print JPEG and full TIFF. The docs don't describe IIIF.
Search
Keyword q with filters such as type, department, cc0, has_image, artists, title, technique, created_after and created_before; skip/limit paging with up to 1,000 records per request. A POST with an image file returns visually similar artworks.
Bulk data
The full dataset as JSON or CSV on GitHub, updated daily.
Best for
High-resolution CC0 files, print products, image similarity search.
curl "https://openaccess-api.clevelandart.org/api/artworks/?q=water%20lilies&has_image=1&cc0&limit=5"

Rijksmuseum API (Rijksmuseum data services)

The Rijksmuseum's current data services are built on Linked Open Data: a Search API, a resolver for persistent identifiers, OAI-PMH and LDES harvesting, data dumps and three IIIF APIs. The Search API needs no key (Rijksmuseum Search API docs). Tutorials written for the older key-based Rijksmuseum API don't match it.

Coverage
Objects in the Rijksmuseum collection. Library and archive records aren't in the Search API.
API key
Not needed.
Limits
Not published in the docs.
Licence
Much of the information and data is public domain or CC0 and can be used commercially. Some material is CC BY 4.0, and photographs of works still in copyright are restricted; the notice on each object says which (information and data policy).
Images
IIIF Image API, IIIF Presentation API and IIIF Change Discovery API (IIIF docs).
Search
Field parameters creator, title, type, material, technique, creationDate (with wildcards such as 16??), description, aboutActor, imageAvailable and objectNumber, in Dutch or English. Each page holds 100 Linked Data identifiers that you resolve for details.
Best for
Dutch art and history, Linked Art and Linked Open Data projects, IIIF.
curl "https://data.rijksmuseum.nl/search/collection?type=painting&creator=Vermeer&imageAvailable=true"

Harvard Art Museums API

The Harvard Art Museums API exposes the data behind the museums' website, with extras such as colours extracted from each image and Elasticsearch aggregations. It needs a key, which you request through a form, and its terms limit it to non-commercial use (Harvard Art Museums API docs).

Coverage
Objects, people, exhibitions, publications and galleries of the Harvard Art Museums, plus vocabularies such as classification, century, culture, medium and technique.
API key
Required on every request as apikey; requested through a form linked from the docs.
Limits
The terms ask you to limit calls to 2,500 per day, not to cache content for more than two weeks, to use the image URLs from the API instead of local copies, and to link back to each object on the museums' website.
Licence
The terms allow non-commercial use only.
Images
IIIF Image API 2.1 and IIIF Presentation manifests. Images of objects with rights restrictions are excluded for most API users.
Search
keyword, q (Elasticsearch query string), classification, century, color, technique, worktype, hasimage and sort=random; up to 100 records per page.
Best for
Research, teaching and non-commercial projects, colour-based browsing.
curl "https://api.harvardartmuseums.org/object?apikey=YOUR_API_KEY&classification=Paintings&hasimage=1&sort=random&size=5"

Europeana API

Europeana aggregates records from libraries, archives and museums across Europe. Its Search API queries all of them at once and lets you filter by the rights each institution has set (Europeana Search API docs).

Coverage
Cultural heritage records of many kinds: art, photography, manuscripts, maps, newspapers and more. theme=art limits a search to the art collection.
API key
Required. Since 28 May 2025 keys live in your Europeana account: a personal key is issued straight away; project keys for services need approval (1 to 5 working days) and have higher limits. Send the key in the X-Api-Key header; the wskey parameter is deprecated (Accessing the APIs).
Limits
No figures are published in these docs. Personal-key limits were reduced step by step until April 2026, and Europeana asks services to use project keys.
Licence
All metadata is CC0. Media follow the rights statement of each item (API FAQ); reusability=open, restricted or permission filters by those rights.
Images
Thumbnails, plus IIIF Presentation API 2.1 and 3, and the IIIF Image API where the providing institution offers it. The IIIF APIs need no key (IIIF APIs).
Search
query with qf refinements, reusability, media, thumbnail, colourpalette, theme and sort (including random order); up to 100 rows per request and cursor-based paging.
Best for
Pan-European projects and apps that must show only openly licensed media.
curl "https://api.europeana.eu/record/v2/search.json?query=Vermeer&theme=art&reusability=open&rows=10" \
  -H "X-Api-Key: YOUR_API_KEY"

Smithsonian API (Smithsonian Open Access)

The Smithsonian's Open Access API covers all of its museums and research units, from the Cooper Hewitt design museum and the Freer and Sackler galleries to the National Air and Space Museum, so it reaches well beyond art. A category search limits queries to art and design (Smithsonian Open Access API docs).

Coverage
More than 11 million metadata records in the Open Access dataset (Smithsonian on GitHub); more than 2.8 million CC0 images at the February 2020 launch (AWS Open Data registry).
API key
Required as api_key. The sign-up form issues an api.data.gov key.
Limits
api.data.gov's default is 1,000 requests per hour per key, though limits can vary by service; the X-RateLimit-Limit response header shows yours (api.data.gov manual).
Licence
Open Access content is CC0. Each media item in a response carries a usage.access value such as CC0.
Images
Links to thumbnail, screen-size and high-resolution JPEG files.
Search
q (free text or fielded), start, rows (up to 1,000), sort (relevancy, id, newest, updated, random), type and row_group; /category/art_design/search for art and design.
Bulk data
Metadata and images on AWS Open Data, updated weekly.
Best for
American art, design and history; CC0 high-resolution images.
curl "https://api.si.edu/openaccess/api/v1.0/category/art_design/search?q=sunflowers&rows=10&api_key=YOUR_API_KEY"

WikiArt, Wikimedia Commons, Wikidata and Artsy

WikiArt API

WikiArt offers a read-only API for artists and paintings (WikiArt API page). The free package allows 4 requests per second, 400 per hour and 10 sessions or IPs per hour; the paid Startup and Premium packages cost $200 and $840 for 12 months and allow 1,000 and 10,000 requests per hour, billed by invoice. You get keys after signing in. The documentation is a Google Doc that its own header marks as obsolete. WikiArt shows both public-domain and copyrighted works, the latter under fair use as low-resolution copies that its terms call unsuitable for commercial use (terms of use). Our WikiArt API alternative guide covers it in detail.

Wikimedia Commons and Wikidata

Wikidata holds structured data about artworks, artists and art movements, all under CC0 (Wikidata licensing). Wikimedia Commons hosts the image files, many of them public-domain paintings, and each file has its own licence and attribution rules (reusing Commons content). You query Wikidata with SPARQL at query.wikidata.org without a key. Each query may run for 60 seconds, a client gets 60 seconds of processing time per minute and 30 failed queries per minute, and an IP may run 5 queries in parallel (Query Service manual). Requests need a descriptive User-Agent header (User-Agent policy). Wikidata is the best source for artist and movement metadata, and the Met's API links its artists to Wikidata.

Artsy API

Artsy's developer site says the public API is being retired and may be taken down without notice. It was limited to public-domain works for educational and other non-commercial purposes (Artsy developers). Don't start a new project on it.

How to choose an art API

Choose a museum's own API when

  • Your project is about one collection. The museum's API has its full catalogue data, its own identifiers and links back to its website.
  • You need CC0 images, for example for prints or merchandise. The Met, Art Institute of Chicago, Cleveland and Smithsonian flag public-domain or CC0 works per object, and the Rijksmuseum states the licence on each object. Art Search doesn't return rights information per image.
  • You need IIIF for deep zoom or a viewer such as Mirador: Art Institute of Chicago, Rijksmuseum, Harvard, and Europeana where the institution provides it.
  • You want to build a dataset or run an analysis. The Met, AIC, Cleveland, Rijksmuseum and Smithsonian publish bulk downloads, and CC0 lets you keep the data. Art Search's terms don't allow storing its data.
  • Your budget is zero and the project is commercial. The keyless CC0 APIs cost nothing; the Art Search free plan is non-commercial.

Choose Art Search when

  • You want one search across many museums instead of integrating and merging several APIs with different schemas, paging rules and licences. (Europeana is the free alternative here, with European collections and a key per project.)
  • Your users search by meaning. A query such as "a lonely figure in a snowy landscape" is matched by meaning, not only by the words in its title. See the semantic art search tutorial in Python.
  • You need filters that work the same for every artwork: creation dates, type, material, technique, origin and image aspect ratio (handy for fitting a layout).
  • You need random artworks for a daily painting, a screensaver or a game. See the Random Artwork endpoint and the painting of the day tutorial.
  • You ship a commercial product and want a plan with commercial use, published quotas and a 99.9% uptime guarantee instead of building on a free API whose terms or capacity may not fit a commercial service (Harvard's API, for example, is non-commercial).

Combine them when

Nothing stops you from using both: a museum API for a featured collection or for CC0 print files, and Art Search for discovery across collections. Wikidata fills in artist and movement data for either.

Sources