Hardcover API vs Big Book API
Hardcover and Big Book API are often compared, but they cover different ground. Hardcover is a reading tracker whose GraphQL API exposes books together with what readers do with them: shelves, lists, reviews and ratings. Big Book API is a REST API for searching books, getting book records and finding similar books, with no user data. This page compares limits, authentication and terms as of November 2026, and says which fits which project.
Disclosure: we make Big Book API, one of the APIs in this comparison. Every other API is described from its own documentation and pricing pages as read on November 22, 2026, plus a few secondary reports that are marked as such; sources are linked below. If something has changed, email [email protected] and we will update the page.
The Hardcover facts on this page come from the source files of its documentation, Getting Started and OAuth, because the rendered docs site was behind a bot check when we read it.
At a glance
| Hardcover API | Big Book API | |
|---|---|---|
| Style | GraphQL, one endpoint | REST, four GET endpoints, JSON |
| Authentication | personal access token or OAuth | API key in x-api-key header or api-key parameter |
| Free use | 5,000 requests/day, 60/minute | 50 requests/day, 1/second, non-commercial, backlink |
| Paid use | Supporter: 50,000/day; commercial tier "building that out right now" | $6, $29 or $99/month for 500, 5,000 or 50,000/day, commercial use |
| Data | books plus reader data: shelves, lists, reviews, ratings, reading journals | 4 million+ English books, authors; no user data |
| Search | 2-second timeout; pattern operators such as _like disabled | natural-language query plus filters |
| Similar books | not documented | up to 100 per book |
| Data-use rules | no training of public or commercial LLMs; user-owned data needs the user's authorization | no storing or reselling; caching up to 1 hour only with written permission |
The Hardcover API
GraphQL, in beta
Hardcover's API lives at https://api.hardcover.app/v1/graphql and is "exactly the same API used by the website, iOS and Android apps". So it reaches the same data the apps show, through queries you shape yourself. The docs are frank about the status: "The API is currently in beta", it is "heavily in flux right now. Anything you build using it could break in the future", and "We may reset tokens without notice while in beta."
Rate limits
| Plan | Requests per day | Burst | Per minute |
|---|---|---|---|
| Free | 5,000 | 10 | 60 |
| Supporter | 50,000 | 15 | 60 |
Each top-level field in a query counts as one request, so a query that asks for three things at the top level uses three. A query with more top-level fields than the burst limit is rejected with HTTP 403 request_exceeds_capacity. Daily limits reset at midnight UTC. The Supporter membership costs $4.99 a month or $49.99 a year according to the US App Store listing; we could not confirm the price on hardcover.app.
Tokens changed in 2026
If you followed a tutorial written before August 2026, parts of it are out of date:
- On August 20, 2026, personal access tokens replaced the old JWT tokens. You choose how long a token stays valid when you create it, and tokens now have permission scopes; older tokens count as
all. - In September 2026 Hardcover added OAuth: the authorization code flow with mandatory PKCE, plus a device grant. Access tokens last one week; refresh tokens last six months and rotate on every use.
- For apps other people sign into, Hardcover recommends OAuth and says "Never ask a user to paste a PAT into your app."
The July 2026 version of the docs said tokens expired after one year and reset on January 1, gave a limit of 60 requests a minute and stated no daily limit, and said the API "should only be used from a code backend".
Query restrictions
- The operators
_like,_nlike,_ilike,_regex,_iregex,_nregex,_niregex,_similarand_nsimilarare disabled, so substring and pattern filters are not available. - Queries time out after 30 seconds, search queries after 2 seconds.
- Browser use is unclear. One line says "Queries are not allowed to run in the browser"; another says "CORS is open, so requests from a web browser will work". Hardcover's 2026 roadmap lists "Client-side API use will be supported". Until that lands, keep tokens on a server.
- Planned for 2026 and not yet in place: a maximum query depth of 3, and queries that "may be limited to your own user data, public data, and user data of users you follow".
User data and LLM training
Two rules matter for commercial projects. First: "By using the API, you agree not to use to train large language models that are made publicly available or used commercially." Second, commercial products and public websites "may not use data owned by users", which covers reviews, ratings, lists, reading journals and similar data, unless the user authorizes access. Aggregate data is allowed if it is attributed to Hardcover. If you republish user-uploaded images, Hardcover advises keeping a DMCA policy. A dedicated commercial tier is not available yet: "We're building that out right now".
Big Book API
Big Book API has four endpoints, all HTTPS GET requests that return JSON: /search-books, /{id}, /{id}/similar and /search-authors. Search parses the query against an ontology, so "books about dogs" also matches books about border collies, and filters by 92 genres, authors, a rating range, publish years, ISBN-13 and OCLC number. Each call costs one request, whatever it returns.
curl -H "x-api-key: $BIG_BOOK_API_KEY" \
"https://api.bigbookapi.com/search-books?query=cozy+mystery+in+a+bookshop&min-rating=0.7&number=10"
curl -H "x-api-key: $BIG_BOOK_API_KEY" "https://api.bigbookapi.com/16384516/similar?number=10" The limits are real too. The catalogue is English-only and has over 4 million books, fewer than the large commercial databases. There is no user data of any kind. The free plan is small: 50 requests a day at 1 per second, for non-commercial use with a backlink, so Hardcover's free tier is far larger. The terms do not allow storing data or reselling it, and caching for up to an hour needs written permission. Paid plans are listed on the pricing page; over the daily quota the free plan returns HTTP 402, and going over the rate limit returns 429.
Which one fits
Hardcover is the better fit when you need reader data: a user's shelves and reading status, lists, reviews and ratings, or social features around them. It is also the better fit for personal tools and hobby projects that need many requests for free, and for apps whose users already track their reading on Hardcover and can sign in with OAuth.
Big Book API is the better fit when you need search by what a book is about, "similar books" recommendations, or a plain REST API you can call with a key from any backend. It suits commercial products that need terms for commercial use today, as long as they work within the storage restrictions.
Using both works: Hardcover, through OAuth, for the user's own shelves, and Big Book API for search and recommendations, with books matched by ISBN-13. Big Book API book records carry identifiers.isbn_13 when the book has one; a book with only an ISBN-10 converts to ISBN-13 in one step. If you mainly need free data you can store, look at Open Library instead; the book API comparison covers it.
Frequently asked questions
Is the Hardcover API free?
Yes, with a Hardcover account: 5,000 requests a day and 60 a minute. Supporter members get 50,000 a day. A commercial tier is still being built.
Is the Hardcover API GraphQL or REST?
GraphQL, at https://api.hardcover.app/v1/graphql. It is the same API the Hardcover website and apps use. Big Book API is REST.
Do Hardcover API tokens expire?
Since August 20, 2026 you set the lifetime of a personal access token when you create it. OAuth access tokens last one week and refresh tokens six months. Hardcover may also reset tokens without notice while the API is in beta.
Can I use Hardcover data in a commercial app?
Not data owned by users, such as reviews, ratings and lists, unless the user authorizes access. Aggregate data is allowed with attribution, and the API may not be used to train public or commercial language models.
Does Big Book API have shelves or reviews like Hardcover?
No. It has book search, book records, similar books and author search. For reader data, use Hardcover or your own database.