Zhihu question answers API

The first 20 answers to a question, by ID or link, in Zhihu's own order: the question as title, full text, author and their followers, upvotes, comments, saves and time posted. One page: there is no next.

GEThttps://api.okdata.online/v1/zhihu/questions/answers/list

Also calledanswers to a zhihu questionzhihu answer list

Price: $0.002; per call · failed calls are free
Returns: A list of Post; the whole list in one answer
If a call fails: Not charged; When a data source fails, the next one is tried; errors and timeouts cost nothing.
Status: preview; It may still change; any change is announced beforehand.

Parameters

Name	Type	Description
`id`	string	The question ID (digits): the number after zhihu.com/question/. Give id or url. e.g. `55043447`
`url`	string	The question link (an answer link under it works too): zhihu.com/question/…. e.g. `https://www.zhihu.com/question/55043447`
`view`	stringcompactstandardfull	How much of each object comes back. full adds platform-specific fields under ext. default `standard`
`fields`	string[]	Fields to keep, by path (author.name), to save tokens. In a query string: comma-separated.

Give exactly one of id, url. cursor, view and fields work the same on every endpoint: how requests work.

Request

cURL

curl -G "https://api.okdata.online/v1/zhihu/questions/answers/list" \
  -H "Authorization: Bearer $OKDATA_API_KEY" \
  --data-urlencode 'id=55043447'

Python

import os
import requests

resp = requests.get(
    "https://api.okdata.online/v1/zhihu/questions/answers/list",
    headers={"Authorization": f"Bearer {os.environ['OKDATA_API_KEY']}"},
    params={"id": "55043447"},
    timeout=60,
)
body = resp.json()
if body["code"] != "ok":
    raise RuntimeError(f"{body['code']}: {body['message']} (charged: {body['charged']})")

for row in body["data"]["items"]:
    print(row["id"], row["metrics"]["likes"], row["text"][:60])
print("cost: $" + body["meta"]["billing"]["amount"])

TypeScript

const url = new URL('https://api.okdata.online/v1/zhihu/questions/answers/list');
url.searchParams.set('id', "55043447");

const res = await fetch(url, {
  headers: {
    Authorization: `Bearer ${process.env.OKDATA_API_KEY}`,
  },
});
const body = await res.json();
if (body.code !== 'ok') throw new Error(`${body.code}: ${body.message}`);

for (const row of body.data.items) {
  console.log(row.id, row.metrics.likes, row.text.slice(0, 60));
}
console.log('cost: $' + body.meta.billing.amount);

MCPtools/call

{
  "name": "call",
  "arguments": {
    "capability": "zhihu.questions.answers.list",
    "id": "55043447"
  }
}

Response

200 · application/jsonmeta.billing.amount = 0.002

{
  "code": "ok",
  "request_id": "0f6c2a8e-4b1d-4e0a-9a51-3c7d2b9e1f20",
  "data": {
    "items": [
      {
        "id": "3283784039",
        "url": "https://www.zhihu.com/question/568657112/answer/3283784039",
        "type": "text",
        "title": "预算 3000 元左右，有哪些不错的露营装备推荐？",
        "text": "先说结论：帐篷和睡袋别省，其他都可以慢慢换。",
        "lang": "zh",
        "published_at": "2026-11-02T09:30:00Z",
        "author": {
          "id": "b32409944256dfd62a963434fcda6aa9",
          "username": "qingzhuang-tubu",
          "name": "轻装徒步",
          "avatar_url": "https://cdn.example.com/avatar/trailkit.jpg",
          "followers": 15700,
          "verified": true
        },
        "metrics": {
          "views": null,
          "likes": 48200,
          "comments": 1312,
          "shares": null,
          "favorites": 9650,
          "reposts": null
        },
        "images": null,
        "hashtags": null
      }
    ]
  },
  "page": {
    "next_cursor": "eyJvIjoyMH0",
    "has_more": true
  },
  "warnings": [],
  "meta": {
    "capability": "zhihu.questions.answers.list",
    "billing": {
      "amount": "0.002",
      "currency": "USD",
      "balance_after": "9.998"
    },
    "unavailable": {
      "metrics.views": "platform",
      "metrics.shares": "platform",
      "metrics.reposts": "platform"
    },
    "cached": false,
    "replayed": false,
    "fetched_at": "2026-11-11T08:00:00Z"
  }
}

Try it

This sends a real call to https://api.okdata.online and is billed as usual: at most $0.002. Failed calls are free. Your key goes only to that address and is never stored on this site.

Known limits

The whole list comes in one answer; there is no next page.
Give exactly one of id, url.
3 fields of Post this endpoint cannot give

Returned fields

data.items is a list of Post.

Field	Type	View	Description
`id`	string	compact	Post ID on the platform.
`url`	string	compact	Link to the post.
`type`	string	compact	What kind of post it is.
`title`	string \| null	compact	Title, where the platform has one (videos, notes, articles).
`text`	string	compact	Caption, body or description as plain text; articles come as Markdown.Long text: view=compact may shorten it.
`lang`	string \| null	standard	Language code, as reported or detected.

Show all 12 fieldsHide the other 6

Field	Type	View	Description
`published_at`	string \| null	compact	When it was published, ISO 8601 in UTC; null only where the source gives no time (wechat_mp.articles.metrics, wechat_channels.videos.metrics, youtube.channels.videos.list with type=short), then listed in meta.unavailable.
`author`	AccountRef	compact	Who posted it.
`metrics`	Metrics	compact	Counts at fetch time; a count the platform hides is null.
`images`	string[] \| null	standard	Image URLs for image posts and notes.
`hashtags`	string[] \| null	standard	Hashtags or topics, without the # sign.
`ext`	object	full	Platform-specific extras. Only returned with view=full.

The view column is the smallest view a field comes back in: compact returns only compact fields, standard adds the standard ones, full returns everything. The default is standard.

What this endpoint cannot give

These fields of Post come back as null here, and meta.unavailable names each one with its reason.

metrics.viewsthe platform does not show it
metrics.sharesthe platform does not show it
metrics.repoststhe platform does not show it

When it is charged

One call costs $0.002 and returns one page of about 20 posts. Collecting 1,000 posts takes about 50 calls, $0.10 in total. Each page is a separate call; failed calls are not charged.

Data comes back$0.002 per call

Bad parameters, errors, timeoutsFree

When	Charge
Most one call costs	$0.002
An empty answer	$0.002
Not found (the platform has no such item)	Free
Bad parameters, errors, timeouts	Free
A retry with the same Idempotency-Key within 24 hours	Charged once

A failed call is free; an empty answer costs $0.002; not found is free.

Prices from price book 1b273c48fea9634e. Every price on this platform →

Common errors

Every error has the same shape and says whether it was charged. The ones this endpoint can return:

HTTP	Code	Meaning	Charged
400	`invalid_params`	A parameter is missing, has the wrong type or value, or the method is wrong. problems lists each one; param names the first.Retry: after a fix	no
401	`unauthorized`	The key is missing, malformed, revoked, disabled or expired (reason says which when known).Retry: no	no
402	`insufficient_balance`	The balance (available) cannot cover the most this call could cost (required).Retry: after a fix	no
404	`not_found`	The post, account or page does not exist or is not public. Endpoints whose charge_when_not_found is above 0 charge this answer (the source billed us for it): the body then says charged: true and amount.Retry: no	no
429	`rate_limited`	The account sent too many requests this minute (reason: rate) or has too many calls running (reason: concurrency).Retry: yes	no
500	`settlement_unknown`	The data came back, but the charge could not be confirmed.Retry: with the same Idempotency-Key	unknown until looked up
502	`upstream_failed`	Every data source for this endpoint failed on this request.Retry: yes	no
503	`capability_paused`	This endpoint is paused right now; nothing ran.Retry: yes	no
504	`timeout`	The data did not come back in time. The hold was released, so nothing was charged.Retry: yes	no

All error codes →

Questions

How much does the Zhihu question answers API cost?

One call costs $0.002 and returns one page of about 20 posts. Collecting 1,000 posts takes about 50 calls, $0.10 in total. Each page is a separate call; failed calls are not charged.

How do I get the next page?

Pass page.next_cursor from the response back as the cursor parameter. When page.has_more is false there is nothing left. Every page is its own call at $0.002.

Do I need a Zhihu account, cookies or a QR login?

No. You send only your okdata API key. The data is public data from Zhihu; nothing is fetched with your own Zhihu account.

Am I charged when a call fails?

No. Every error response says charged: false, except a timeout, which says charged: null until the outcome is known. Resend with the same Idempotency-Key and you are never charged twice.

What if Zhihu does not show a field?

The field is null, and meta.unavailable names it with the reason: platform (the platform never shows it) or capability (we cannot fetch it yet).

Use it in your AI

Paste this into Claude, ChatGPT, Cursor or any agent that can open a web page. It reads this endpoint’s page and calls it with your key.

Read https://okdata.online/apis/zhihu/questions/answers/list.md, then call this endpoint for me through okdata; my key is in the OKDATA_API_KEY environment variable. Use Zhihu question answers and list the five posts with the most likes, with links.

With the MCP server connected, just ask: Use Zhihu question answers and list the five posts with the most likes, with links.

Connect your agent → MCP →

All 15 Zhihu endpoints

Main Zhihu endpoints

Zhihu API overview →Zhihu API pricing →

Names and ID

Name: Zhihu question answers API
Short name: Zhihu question answers
ID: zhihu.questions.answers.list
Endpoint: GET /v1/zhihu/questions/answers/list
Category: Social media
Also called: answers to a zhihu question, zhihu answer list
Returns: Post[]: 12 fields