My Zhihu scraper stopped working: what now? (2026)
2 routes need no scraper: Zhihu's own data platform, with search, a hot list and question answers under its own quotas, or a per-call data API. On the second, zhihu.posts.search gives about 18 answers and articles a page and zhihu.questions.answers.list the first 20 answers to a question with their text, each at $0.0020, with one key and no account or cookie of yours. DataXAPI reads Zhihu with 16 endpoints, from $0.0020 a call; a failed call is not charged.
Updated 2026-11-04
Zhihu's own platform, and a data API
What Zhihu's column says is from its documents as read on 2026-11-04, and its free quota is called a limited-time offer.
| Job | Zhihu data platform | DataXAPI |
|---|---|---|
| Search | zhihu_search: up to 10 results, and it says HasMore is always false for now | zhihu.posts.search: about 18 a page, with further pages; type picks answers or articles, sort orders by relevance, newest or upvotes |
| A question's answers | question_answers: 1 to 50 a request, with paging; each is a summary, not the whole answer | zhihu.questions.answers.list: the first 20 in Zhihu's order, one page only, with text, author, upvotes, comment and save counts |
| Comments under an answer | Its documents list comments on your own creations | zhihu.answers.comments.list: about 10 a page, by the answer's id |
| Quota and price | Its home page says 5,000 free calls a day on signing up. Past the free quota, search and question answers are 20 yuan per 1,000. Question answers share a default of 100 a day | $0.0020 a call, from your balance; a failed call is not charged |
The rest of what a scraper usually did
zhihu.questions.get reads a question: title, details, views; view=full adds ext.answers and ext.followers. zhihu.answers.get reads one answer's text and zhihu.articles.get one article with its counts. zhihu.users.answers.list gives an author's answers, newest first, by the handle in their profile's address. zhihu.trends.list gives the hot list, about 30 questions.
You send your key and the question. No signature headers to produce, and no cookie or account of yours.
What a study comes to
The three lines together come to $0.82. On the command line: dataxapi call zhihu.posts.search query=咖啡机推荐 type=answer --pages 3 --max-spend 0.006 --out answers.json.
| Job | Calls | Comes to |
|---|---|---|
| 20 keywords, 3 pages each: about 1,080 answers and articles | 60 × $0.0020 | $0.12 |
| The answers to 50 questions: about 1,000 | 50 × $0.0020 | $0.10 |
| 3 pages of comments under 100 answers: about 3,000 | 300 × $0.0020 | $0.60 |
Good to know
A question's answers stop at the first 20, so if you need every answer to a big question, this does not do it. Upvotes come in metrics.likes. Search results and answer lists have no read counts. It reads public data only and cannot post. How to keep a scraper from being blocked is not something this page covers.
Endpoints
Each with its price; every page has the parameters, the fields and a recorded response.
Sources
What this page says of other services is from their own pages and documents, read on 2026-11-04. Prices and features change: check the source before you decide.