WeChat Official Account Data API in 2026: Official, GitHub or Pay-per-Call

WeChat Official Account API in 2026: the official API reads only your own account, top GitHub tools stopped, and pay-per-call starts at $5.56 per 1,000.

Hand-drawn illustration: on a paper-white background, a tall, uneven pile of folded white article sheets stands on the left. One large sheet has been pulled out to the right; along its bottom edge sit three small blocks, the middle one yellow. A magnifier rests on it beside a speech bubble and a few yellow coins.

Yes, you can get WeChat Official Account data through an API, and the first question is whose account you want to read. There are three kinds of routes. The first route is the official server API of the WeChat Official Accounts Platform. The second route is an open-source scraper from GitHub.

The third route is a pay-per-call data API. We compare two pay-per-call vendors: EveryInfra, which is our own API, and TikHub. The EveryInfra numbers come from live calls at three times on November 22, 2026. The TikHub numbers come from its public price list; we did not call it.

Each route fits a different job. Use the official API to manage your own account. It can publish posts, read the traffic sources of your own posts and moderate comments. The official API has no endpoint that reads other accounts. Do you want to read public posts, read counts and comments from other accounts in bulk? Then choose between open-source tools and pay-per-call APIs.

Each of the three routes changed in 2026. The official page for the published-post list was last edited on November 11, 2026. Its scope table says "verified only" for Official Accounts. It also says that from July 2025, WeChat takes back this API from personal accounts and from unverified company accounts. On GitHub, the most-starred export tool, wechat-article-exporter (13,172 stars), stopped on July 30, 2026. Its website closed in October.

The second, wewe-rss (9,615 stars), is archived. On the pay-per-call side, the EveryInfra catalog lists 13 Official Account capabilities, and 10 are available. They cost $5.56 to $16.00 per 1,000 calls. Billing is per call, not per row. At 12:56 SGT we read the metrics of one health article. The reply gave read_count 100001, 122 "Wow" taps and 3,469 shares, for $0.005556.

Generated illustration: at the top, a teal tray shaped like a speech bubble holds a pile of article pages. Three teal paths lead down to three things. On the left, an iron gate and a mailbox with a yellow key that opens only its own box. In the middle, a messy workbench with a laptop, tangled cables, a QR code card and a yellow warning sign. On the right, a pipe with a meter showing a stack of coins, which delivers a neat stack of articles.
Each route leads to a different thing: the official API opens your own mailbox, open source gives you a workbench to maintain, and a pay-per-call API counts calls and delivers clean public posts. Generated illustration based on our comparison of the three routes

What is a WeChat Official Account, and why is its data hard to get?

A WeChat Official Account is a publishing account inside WeChat. Brands, media, schools and government offices use it to push articles to followers. The articles live at mp.weixin.qq.com web addresses. You can open one in a normal browser. But the read count, the "Wow" count and the comments load later, through calls that expect the WeChat app. For this reason, the text of one article is easy to get, but the numbers around it are not.

An Official Account is not a personal WeChat account. It is also not WeChat Channels, which is the short video product. WeChat has a separate developer program for mini programs and another for WeChat Pay. Here, "Official Account data" means four things: the list of articles from one account, the article text, the article metrics (reads, likes, "Wow", shares, saves, comment count), and the public comments.

There is one more difference for readers outside mainland China. The official documents are in Chinese. Account verification needs a mainland company or an overseas company that WeChat accepts, and it has a yearly fee. Many open-source tools assume that you run a WeChat client with a Chinese phone number. A pay-per-call API avoids these steps, because you only send an HTTP request with a key.

WeChat Official Account API comparison: one table for all three routes

The official column comes from the WeChat Open Docs, which we read page by page on November 22, 2026. The open-source column comes from the README files of 10 popular projects on GitHub, which we read at 12:44 SGT on the same day. For the two pay-per-call columns, we read EveryInfra prices from its public catalog at 12:37:42 SGT. We read TikHub prices from its public price list, which needs no key, at 12:47:43 SGT, and from its FAQ. For a line-by-line price comparison, read our TikHub alternatives page and TikHub pricing explained.

ItemOfficial API (WeChat Official Accounts Platform)Open-source scrapers on GitHubTikHubEveryInfra
Whose dataOnly your own account (or accounts that authorize your third-party platform)Any public post, through the account back office search, WeChat Read or the WeChat desktop appPublic posts, accounts, commentsPublic posts, accounts, comments, metrics
Account and paperworkCompany account with WeChat verification; personal accounts lost the publish API from July 2025Most need a QR login to an Official Account or a WeChat Read accountSign upSign up
Cookie or sessionAppSecret for an access_token; your server IP must be on an allowlistYes; scan again when it expiresPost web address or gh_ ID; no cookie from youNo
Billing unitThe API is free; verification has a feeFree; you pay for servers, accounts and upkeepPer request, charged only on HTTP 200Per call, not per row; failures and empty results are free
Price per 1,000 callsNoneNone$10.00 (23 entries), $1.00 (1 entry)$5.56 to $16.00
Daily cap"List published posts": 100 a day for a new accountUntil the account gets blocked10 requests per second per endpointPer key; the test key here allows 10 a minute
MCPCommunity MCP servers for your own account, still with your AppSecretSome offer RSS or an APIHas MCPRemote MCP, tool names everyinfra_*
Best forYour own account: publishing, traffic sources, comment moderationSmall one-off jobs, if you can fix breaksTeams that already use its other platformsBulk reads of other accounts' posts, reads and comments

Both pay-per-call vendors bill per call, so first convert prices to one unit. For article text or metrics, one request gives one result, so compare the price per 1,000 calls. For search, comments and article lists, one call gives many rows. The price per 1,000 rows then depends on the number of rows per call. EveryInfra wechat_oa/article_search costs $0.008 per call, and one call can return up to 400 rows.

At 12:38 SGT we searched for "camping" with default settings. We got 18 rows, which is $0.444 per 1,000 rows. If a keyword is popular enough to fill 400 rows, the cost falls to $0.020 per 1,000 rows. The TikHub FAQ says that its WeChat search has no pages: to get more results, you send the same request again and accept some repeats. We did not test how many rows it returns, so the table gives no per-row price for TikHub.

Horizontal bar chart, USD per 1,000 calls: article body, account owner info and metrics at 5.56 dollars each; comments and comment replies at 6.76 dollars each; article search, account profile, past articles and today's articles at 8.00 dollars each; account search at 16.00 dollars. Teal marks the five capabilities we tested: article body, metrics, comments, article search and past articles. The catalog lists 13; three paused ones are not shown.

Does WeChat have an official API for Official Accounts?

Yes. The WeChat Official Accounts Platform gives each account a full server API. The documents are at developers.weixin.qq.com. They are now split into one set for Official Accounts (the old "subscription accounts") and one set for Service Accounts. To call the API, you must own an account. You scan a QR code on the WeChat developer platform and get the AppID and the AppSecret.

You add your server IP to an allowlist. Then you call Get access token. The API guide says that a token lasts 2 hours, and a new token makes the old one invalid. It tells you to refresh tokens from one central server. A server IP that is not on the allowlist gets error 61004.

Three groups of endpoints give data, and all three work only on your own account. The first group is publishing. List published posts (/cgi-bin/freepublish/batchget) returns 1 to 20 posts per call, with title, author, summary, HTML body and a deleted flag. Get a published post reads one post by article_id. The second group is statistics. The statistics guide says these endpoints are open to all verified accounts.

The old daily article summary is no longer maintained. The new daily read data splits readers by source: account messages, chats, Moments, recommendations, search and others. It has data only from November 1, 2025, and each call covers one day. The third group is comments. List comments reads comments by the msg_data_id from a mass send. It refuses 50 or more per call, and it returns each commenter's openid.

The scope table on these pages says the same thing: "verified only" for Official Accounts. That means only verified company accounts can call them. The publishing pages add a note. From July 2025, WeChat takes back the API from personal accounts, from unverified company accounts and from accounts that cannot be verified. Here is the original text, captured on November 22, 2026.

Screenshot of the WeChat Open Docs page for listing published posts. Section 7, Scope, shows a table: the Official Account column reads verified only, and the Service Account column has a check mark. Three notes explain that verified only means verified company accounts. The last line, in a teal box, says that from July 2025 personal accounts, unverified company accounts and accounts that cannot be verified lose access.
Scope of the publish API: for Official Accounts, only verified company accounts can call it, and personal accounts lost access from July 2025. WeChat Open Docs, List published posts, captured Nov 22, 2026
Official endpointPathWhat you getWhat you do not getScope for Official Accounts
List published posts/cgi-bin/freepublish/batchgetTitle, author, summary, body and deleted flag of your own postsOther accounts' posts; mass-sent posts with notification (a pinned forum reply says there is no endpoint for them)Verified only
Daily read data/datacube/getarticlereadReaders per post, split by source, for your own accountData before Nov 1, 2025; other accounts' readsVerified only
Daily article summary (old)/datacube/getarticlesummaryReads, shares and savesNo longer maintained; posts with a total under 3 are not countedVerified only
List comments/cgi-bin/comment/listComments, featured flag, author replies, openid for your own postsOther accounts' comments; 50 or more per call is refusedVerified only
Get access token/cgi-bin/tokenaccess_token, valid 2 hoursError 61004 when the IP is not on the allowlistAll accounts

The note about mass-sent posts comes from a pinned reply on the WeChat developer forum on June 27, 2024. It says that the list endpoint returns only posts published without a push notification. So for an account that pushes most posts to followers, the official API may not list even its own full history. You then copy links by hand from the back office.

How many official calls can you make each day?

The API limits page gives daily caps for a new account. Get access token: 2,000. List drafts: 1,000. List followers: 500. List published posts and get a published post by article_id: 100 each.

One call lists up to 20 posts. So a new account can list up to 2,000 of its own published posts a day. Most accounts can export their full history in one day. Mass-sent posts with notification are not in this endpoint. Copy those links by hand.

A third-party platform that calls for an account uses that account's own quota. For a team that manages its own account, 100 calls a day is enough. For monitoring other accounts, the number does not matter, because the API returns only your own posts.

Horizontal bar chart, log scale with 10x per step, calls per day: get access_token 2,000; list drafts 1,000; list followers 500; list published posts 100; get a published post by article_id 100; mass send 100. The two publish bars are teal.

When the official route is the right one

For your own account, the official API gives more than any third party. For traffic sources, getarticleread tells you how many readers came from account messages, from Moments or from recommendations. Only the account owner can see this split. No tool that reads public pages has it.

For comments, comment/list returns the commenter's openid, which you can match with your own follower data. A third-party API sees only the public nickname and region. Drafts, publishing and deleting are write actions, and only the official API can do them.

If you do not want to write the code, community MCP servers exist for your own account.

One example is xwang152-jack/wechat-official-account-mcp. Its README says it stores the AppSecret encrypted on your machine. It still calls the official API, so it still reads only your own account.

WeChat article scrapers on GitHub: do they still work?

We searched GitHub for WeChat article scrapers and sorted by stars. We picked 10 projects that read Official Account articles. At 12:44 SGT on November 22, 2026, we read their stars, last commit and license with the GitHub REST API. Then we read each README to see what login it needs. None of the 10 can list the full history of an account, with read counts, without a login or a WeChat session.

RepositoryStarsLast commit (UTC)LicenseHow it gets data, and login needs (from the README)Status
wechat-article/wechat-article-exporter13,1722026-11-09MITUses the article search inside the Official Account editor; QR login to an account; reads and comments need captured credentialsStopped on 2026-07-30; website closed in October
cooderl/wewe-rss9,6152026-03-20MITUses WeChat Read; QR login; the README warns that frequent adds get the account blocked for 24 hoursArchived
chyroc/WechatSogou6,4112026-10-23Apache-2.0Sogou WeChat search; no login; captcha retry setting; optional proxyStill has commits
rachelos/we-mp-rss5,2252026-11-22CustomSelf-hosted service; QR authorization; scheduled updates of followed accountsActive
wnma3mz/wechat_articles_spider3,4862024-04-18Apache-2.0Cookie and token from the WeChat desktop app by hand; read parameters expire in 4 hoursNo commits for over two years
bowenpay/wechat-spider3,4022021-08-10NoneSogou WeChatREADME says it stopped after Sogou removed the feature on Oct 29, 2019
jackwener/wechat-article-to-markdown1,0912026-03-22NoneOne article link to Markdown; no loginSingle posts only
yeximm/Access_wechat_article7492026-08-23CustomWindows desktop tool; a MITM proxy reads the WeChat desktop app trafficUpdated
wonderfulsuccess/weixin_crawler4942024-02-27NoneProxies the WeChat client and WeChat Read; reads, likes, "Wow"Not updated
Moore-developers/moore-wechat-article-downloader2972026-08-01MITShort session from the WeChat desktop app; body, featured comments, metricsUpdated
Horizontal bar chart of stars and last commit date for 10 WeChat article repositories: wechat-article-exporter 13,172 stars, Nov 9, 2026, light gray for stopped; wewe-rss 9,615, Mar 20, 2026, light gray for archived; WechatSogou 6,411, Oct 23, 2026, teal for no login; we-mp-rss 5,225, Nov 22, 2026, dark gray for QR login; wechat_articles_spider 3,486, Apr 18, 2024, dark gray; wechat-spider 3,402, Aug 10, 2021, light gray for abandoned; wechat-article-to-markdown 1,091, Mar 22, 2026, teal; Access_wechat_article 749, Aug 23, 2026, dark gray; weixin_crawler 494, Feb 27, 2024, dark gray; moore-wechat-article-downloader 297, Aug 1, 2026, dark gray.

The project with the most stars changed the most this year. The README of wechat-article-exporter explains its method. In the Official Account editor, you can search other accounts' posts and insert a link. The tool used this search to list all posts of any account.

On July 30, 2026, the author announced the end of the project in issue #200. The author wrote that the one-click sync of Official Account posts "no longer exists". The repository became read-only, with no new issues or pull requests. A note at the top of the README adds that the website closed in October 2026.

Screenshot of the README of the GitHub repository wechat-article-exporter. Under the title are four badges: Star 13k, Fork 2.3k, License MIT and version v2.3.21. A teal box marks a yellow Warning notice. It says the project stopped on 2026-07-30, and lists four points: the repository becomes a read-only archive, it takes no new issues or pull requests, the website closed in October 2026, and posts you already synced stay on your own device. A middle paragraph that explains the reason is left out of the capture.
The most-starred WeChat export tool stopped on July 30, 2026, and its website closed in October. We left out a middle paragraph; issue #200 has the full text. GitHub, wechat-article/wechat-article-exporter README, captured Nov 22, 2026

The second project, wewe-rss, used a different path: the Official Account subscriptions inside WeChat Read. Its README has an account state called "today's penalty box", which means that the account is blocked for a day. The repository has been archived since March 20, 2026. The fourth project, we-mp-rss, is still active. Its last commit was on the day we read it (November 22, 2026, 04:22 UTC).

Five new issues opened in the last 30 days. In issue #485 (October 30, 2026), a user reports a crypto-mining process and an unknown SSH key inside a self-hosted we-mp-rss container. The reporter also writes that the entry point is unknown, and that the cause may not be in the project code. In issue #489 (November 19, 2026), a user asks why the WeChat authorization QR code returns 404. If you host such a service yourself, do not put its admin port on the public internet.

Licenses decide whether you can reuse the code. Three of the 10 use MIT (wechat-article-exporter, wewe-rss, Moore). Two use Apache-2.0 (WechatSogou, wechat_articles_spider). The we-mp-rss and Access_wechat_article projects use custom terms that GitHub cannot classify. Three have no license file: wechat-spider, wechat-article-to-markdown and weixin_crawler.

By default, code without a license is for reading only, not for copying into your product. Read custom terms line by line. The we-mp-rss settings even include a license encryption key, so ask the author before commercial use. Open source is fine for learning and one-off research. For a long-term company pipeline, check three things: the license, the maintenance status and the account risk.

Four methods in open source, and where each one breaks

The 10 projects use four methods, and each method has a known failure point. The table maps each method to its projects and to the failure that the projects themselves describe.

MethodProjectsWhat you getWhere it breaks, per the project itself
Article search in the Official Account back officewechat-article-exporter, we-mp-rssFull post lists of any accountThe exporter's author says in issue #200 that WeChat closed the core interface it used
WeChat Read subscriptionswewe-rssNew posts as RSSIts README describes blocks and a 24-hour wait
Sogou WeChat searchWechatSogou, bowenpay/wechat-spiderSearch results, no loginNo full account history; frequent requests trigger a captcha; wechat-spider stopped when Sogou removed the feature in 2019
Session or captured traffic from the WeChat desktop appwechat_articles_spider, Access_wechat_article, MooreReads, likes and "Wow" countsParameters expire in a few hours; someone must keep the WeChat client open

Only the fourth method gets reads, likes and "Wow" counts, and it is also the method with the most upkeep. Access_wechat_article shows the cost in its own tracker. Issue #33, opened on November 15, 2026, says that the tool installs and runs without errors. But its window stays blank.

Single article bodies are the exception. A WeChat article page is a public web page. The jackwener/wechat-article-to-markdown tool takes one mp.weixin.qq.com/s/... link and converts it to Markdown. A Juejin post from April 2026 compared six ways to get articles. The author found that a plain request often gets a slider captcha.

With the User-Agent of the WeChat in-app browser, the full HTML comes back. The author also warns to keep the rate low, because WeChat can change its rules at any time. You can get the body this way. You cannot get reads and comments this way, because they are not in the HTML. The page asks WeChat for them after it opens, with identity parameters from the WeChat client.

YouTube demos follow the same methods. At 12:49 SGT we ran youtube/search for "WeChat Official Account scraper" in Chinese. The most-viewed of the top 10 results was an n8n workflow tutorial from July 2025, with 11,275 views. It gets the article list from a paid data service (the video says 0.06 yuan per call, about $0.009 at 6.69 yuan per dollar, from the two currencies in the 12:37 SGT catalog).

An HTTP node then fetches each article, strips the HTML tags and saves the text to Notion. We read 20 comments. One viewer asks about read counts and follower counts. One says that the fetch still fails after setting a User-Agent. One asks why the HTTP node never returns. Lists and bodies are easy; metrics and stability are hard.

Is there a ready-made WeChat Official Account API? How pay-per-call vendors compare

Yes. A pay-per-call API moves the login, the session, the parameter refresh and the retry after a block to the vendor. You send one HTTP request. Here we compare EveryInfra and TikHub on four things. They are the billing unit, the price per 1,000 calls, the cost of failures and the need for your own account.

At 12:37:42 SGT on November 22, 2026, the EveryInfra catalog had 13 Official Account capabilities, and 10 were available. They are article search, article body, account search, account profile, account owner info, past articles, today's articles, comments, comment replies and article metrics. Mini program search, search suggestions and WeChat Index were paused. The 10 available capabilities have only four prices. Article body, owner info and metrics cost $0.005556 per call. Comments and replies cost $0.006764.

Article search, account profile, past articles and today's articles cost $0.008. Account search costs $0.016. Each capability page says that billing is per call, not per row, and that failures and empty results are free. For example, the comments page says one call can return up to 260 rows, and the past articles page says up to 160.

Grouped horizontal bar chart, USD per 1,000 rows. Article search, default: our test 18 rows, 0.444 dollars; at the 400-row cap, 0.020 dollars. Article search, LATEST: our test 20 rows, 0.400 dollars; at the cap, 0.020 dollars. Comments: our test 20 rows, 0.338 dollars; at the 260-row cap, 0.026 dollars. Past articles: our test 8 rows, 1.00 dollar; at the 160-row cap, 0.050 dollars.

Because the price is per call, more rows per call means a lower price per 1,000 rows. Our tests used small limits. At 12:46 SGT we read the comments of one article with limit 20. We got 20 rows for $0.006764, which is $0.338 per 1,000 rows.

If the same article had 260 readable comments, one call would cost $0.026 per 1,000 rows. The real yield depends on the comment count and on the keyword. The "cap" bars in the chart show an upper bound, not a measured number.

At 12:47:43 SGT, the TikHub public price list had 24 Official Account entries: 12 in wechat_mp/v2 and 12 in wechat_mp/web. Twenty-three cost $0.01 per call and cannot use the sign-up credit or volume discounts. One old web endpoint, which returns the article body as JSON, costs $0.001 per call. The TikHub FAQ says that the old Official Account and Channels series were retired on June 13, 2026, and rebuilt as the v2 series.

Article endpoints take the article web address, and account endpoints take the gh_ user name. TikHub charges only successful requests, which means HTTP 200. We did not call TikHub, so we cannot tell which old web entries still work.

The evidence in the table below is not equal on both sides. The TikHub column has only public prices. It cannot show how many rows TikHub returns, whether the fields are right, or what a failure looks like. The EveryInfra column has catalog data and live calls. The live test also lists our five failed calls.

Pay-per-call itemTikHub (public price list, not tested)EveryInfra (catalog plus live test)
Billing unitPer request, charged on HTTP 200Per call; failures and empty results are free
Article body$0.01 per call (v2); old web JSON version $0.001$0.005556 per call; our test returned 3,578 characters and 27 images
Article metrics$0.01 per call$0.005556 per call; our test returned 6 numbers: reads, likes, "Wow", shares, saves, comments
Comments$0.01 per call$0.006764 per call; up to 260 rows
Past articles$0.01 per call, by gh_ user name$0.008 per call; up to 160 rows; WeChat ID, gh_ ID or article link (see the live test)
Free credit$0.05 at sign-up; these Official Account endpoints cannot use itSee the pricing page
Top-up feeAlipay 2.6% + $0.30; PayPal and USDT 5% + $0.30See the pricing page
MCPHas MCPRemote MCP

In AI tools that support remote MCP, such as Claude or Cursor, you can add EveryInfra as an MCP server. The model then calls everyinfra_call_api with the platform wechat_oa. Keep the key in an environment variable, and put only the variable name in the config file:

json
{
  "mcpServers": {
    "everyinfra": {
      "type": "http",
      "url": "https://api.everyinfra.com/mcp",
      "headers": { "Authorization": "Bearer ${EVERYINFRA_API_KEY}" }
    }
  }
}

Fields: what the API returns, and which numbers need care

The EveryInfra catalog lists response_fields for each capability. For the five capabilities we tested, the field counts come from the catalog that we read at 12:37 SGT. The live reply is what counts, and the catalog notes say which fields can be missing in some replies.

CapabilityFieldsCommon fieldsMissing fields noted in the catalog
wechat_oa/article_search10title, summary, url, account_name, posted_atkind can be missing; on some requests id and relative_time too
wechat_oa/article_detail16content_text, content_html, author, account_ghid, is_original, ip_regiondigest and hashtags can be missing
wechat_oa/article_metrics7read_count, like_count, watching_count, share_count, collect_count, comment_countNone
wechat_oa/article_comments11text, like_count, reply_count, is_top, province, posted_atNone
wechat_oa/account_articles12title, url, posted_at, position, is_originalaccount_id and is_deleted can be missing

Start with links. With default settings, the url from article search is a temporary link with signature and timestamp. All 18 rows from our 12:38 search were of this type. With "category":"LATEST" at 12:56, all 20 rows had links with __biz, mid, idx and sn.

The body, comments and metrics need the article's own link, the catalog says. That is a /s/<id> short link, or a link with __biz, mid, idx and sn. To go from a default search to the body, look up the account from account_name, then call past articles for permanent links. By the catalog rule, LATEST links can go straight to the body, comments and metrics, but we did not test that.

Read counts need care too. When read_count is a number such as 100001, we read it as the "100k+" label on the WeChat page. It means "more than 100,000". Only the owner can see the exact number in the back office.

Do not misread position. It is the slot inside one push (the lead post is 1). It is not a ranking by reads.

Which WeChat Official Account API should you use? Choose by job

JobRouteWhyWatch out
Weekly report for your own accounts, with traffic sourcesOfficial statistics APIOnly the owner gets reads split by sourceNeeds a verified company account; data starts Nov 1, 2025
Moderate, feature and reply to comments on your own postsOfficial comment APIHas openid; can write backFewer than 50 per call
Watch 50 competitor accounts every dayPay-per-call APIThe official API cannot read other accounts; open source needs QR accountsPast articles: up to 160 rows per call
Track the newest posts on a keywordPay-per-call APIArticle search can sort by newestSearch returns temporary links; swap for permanent links before reading bodies
Study one viral post: reads, likes, commentsPay-per-call APIMetrics return 6 numbers in one call"100k+" means more than 100,000
Save a few dozen bodies for a knowledge baseSingle-post open-source tool, or copy by handSmall volume; text onlyKeep the rate low
Academic archive of a full account historyAsk the owner, or use a public datasetBetter for consent and completenessCopying a post does not give you its copyright
Column chart, 30 days in US dollars, EveryInfra catalog price times calls, assuming 50 accounts post 100 articles a day in total: metrics for 100 new posts a day 16.67 dollars; bodies for 100 new posts a day 16.67 dollars; article lists for 50 accounts a day 12.00 dollars; comments for 20 posts a day 4.06 dollars; latest search for 10 keywords a day 2.40 dollars; total 51.79 dollars. The first two columns are teal.

Here is a monthly bill for "watch 50 competitor accounts". Assume the 50 accounts post 100 articles a day in total. One article list per account per day is 50 calls at $0.008, or $12.00 for 30 days. Metrics once per new post is 100 calls at $0.005556, or $16.67. The body once per post is also $16.67. Comments for 20 selected posts cost $4.06.

One latest search for each of 10 keywords per day costs $2.40. The total is $51.79 a month. Metrics and bodies are the two largest items. If you need only titles and read counts, skip the bodies and save about one third. One metrics read per post misses later growth. To see how reads grow, read metrics on several days; each extra round adds $16.67 for 30 days.

The official API cannot do this job. Open source can, but its cost is elsewhere. To search 50 accounts, you need an Official Account with a QR login. The wewe-rss README already warns that frequent adds get an account blocked. For read counts, someone keeps the WeChat desktop app open and captures traffic, and the parameters expire in hours. Our Douyin data API comparison looks at the same question for Douyin. For Official Accounts, the open-source options are narrower, because the two most-starred projects have stopped.

A live test: body, comments and metrics of a "100k+" post

On November 22, 2026, from 12:38 to 12:56 SGT, we sent 13 requests to the EveryInfra Official Account capabilities. We chose a post from the health account Dingxiang Life Research, published on July 31, 2026 (__biz=MzA4MTg0MjA5OQ==, gh_038cf5c345ed). It is also the sample post on our capability pages. The request looks like this, with the key read from an environment variable:

bash
curl -s https://api.everyinfra.com/api/v1/social \
  -H "Authorization: Bearer $EVERYINFRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"platform":"wechat_oa","action":"article_metrics",
       "params":{"url":"https://mp.weixin.qq.com/s?__biz=MzA4MTg0MjA5OQ==&mid=2655362972&idx=1&sn=b56fabd8e4c53259498100f05f20dbb0"}}'
# reply at 12:56:01 SGT: read_count 100001, like_count 1043, watching_count 122,
# share_count 3469, collect_count 358, comment_count 40; charged_usd 0.005556

At 12:46:02 SGT we read the body. The reply came in 4.45 seconds. The content_text had 3,578 characters, the post had 27 images, ip_region was Zhejiang and is_original was false. The charge was $0.005556.

The reply had 15 of the 16 catalog fields; hashtags was missing, which the catalog allows. The author field held a pen name, which differs from the account name, so count posts per account with account_name or account_ghid. The digest was only two characters long, so use the body for summaries. The post went out at 03:55 UTC, and the first comment came three minutes later. At 12:46:15 SGT we read the comments with limit 20. We got 20 rows for $0.006764.

The top comment had 1,056 likes, and the 20 comments had 2,817 likes in total. Eight comments had replies, 20 replies in total.

The commenters came from 10 provinces, and Guangdong had the most, with 7. All 20 comments came within two hours after the post went out. The metrics show comment_count 40. We set limit to 20, so we did not read the rest. Billing is per call, so limit 40 would have read all of them for the same $0.006764.

For topic research, the comments show what readers want next. One comment with 235 likes asks for an antiperspirant review, and another asks which home hair-removal device to buy. Likes are concentrated: the top 3 comments have 1,586 of the 2,817 likes, or 56%. None of the 20 has is_top set. Reading the top dozen comments by likes is usually enough; you do not need all 260 for every post.

At 12:56:13 SGT we called past articles with the account's gh_ ID. We got 8 rows in 2.29 seconds for $0.008. The account pushed once a day on each of the last three days, with 3 posts per push (position 1 to 3). The newest push was at 04:14 UTC on November 22, 2026. Each url was a permanent link with __biz, mid, idx and sn. You can pass it straight to the body, comments and metrics capabilities.

In Beijing time, the three pushes went out on November 20, 2026 (12:01), November 21, 2026 (11:48) and November 22, 2026 (12:14). The 3 posts of one push share the same posted_at to the second, so group by posted_at and sort by position to rebuild each push. For an account like this, one list call in the afternoon catches the day's push, and limit 8 covers three days.

The 11 calls that reached the WeChat capabilities are listed below in time order (SGT, November 22, 2026). The two HTTP 429 replies at 12:45:54 and 12:46:51 are left out; they came back in under a second and cost nothing.

Time (SGT)Capability and inputHTTPTime to replyRows or resultCharged
12:38:41article_search, "camping", default order20075.4 s18 rows, 2021 to 2024$0.008
12:40:11account_articles, 36Kr by WeChat ID50312.2 sNone$0
12:44:02account_articles, same input, retry50315.2 sNone$0
12:44:21account_search, 36Kr by name50314.4 sNone$0
12:46:02article_detail, permanent link2004.5 s3,578 characters, 27 images$0.005556
12:46:15article_comments, limit 202004.3 s20 comments$0.006764
12:47:01article_detail, temporary search link (another account's camping post)5037.6 sNone$0
12:47:19account_articles, 36Kr by name50313.9 sNone; reply suggests a ticket$0
12:56:01article_metrics, permanent link2004.5 sSix numbers$0.005556
12:56:13account_articles, Dingxiang Life Research by gh_ ID2002.3 s8 posts with permanent links$0.008
12:56:24article_search, LATEST2009.3 s20 rows, Nov 18 to 22$0.008
Horizontal bar chart of response time and charge for 11 WeChat test calls: 12:38:41 article search 75.4 seconds, 0.008 dollars; 12:40:11 and 12:44:02 past articles, 503, no charge; 12:44:21 account search, 503, no charge; 12:46:02 article body 4.5 seconds, 0.005556 dollars; 12:46:15 comments 4.3 seconds, 0.006764 dollars; 12:47:01 body with a temporary link, 503, no charge; 12:47:19 past articles, 503, no charge; 12:56:01 metrics 4.5 seconds, 0.005556 dollars; 12:56:13 past articles 2.3 seconds, 0.008 dollars; 12:56:24 article search LATEST 9.3 seconds, 0.008 dollars. Two HTTP 429 replies are not shown.

At 12:38:41 SGT we searched for "camping" with default settings. After 75.4 seconds we got 18 rows. Their dates ran from June 12, 2021 to April 22, 2024, and 13 were from 2022. All were older than 2024. At 12:56:24 SGT we searched again with "category":"LATEST".

In 9.3 seconds we got 20 rows. All were posted between November 18, 2026 and November 22, 2026, by 19 different accounts. The default order favors relevant old posts. For monitoring, always pass LATEST. The default call was also much slower. When a call comes near the 85-second sync limit, add "mode":"async" to get a job_id, and poll it.

ItemDefault (12:38:41)`"category":"LATEST"` (12:56:24)
Time to reply75.4 s9.3 s
Rows1820
Post dates2 in 2021, 13 in 2022, 2 in 2023, 1 in 2024Nov 18: 2, Nov 19: 3, Nov 20: 7, Nov 21: 5, Nov 22: 3 (2026)
Different accounts1419
url formAll 18 with signature and timestampAll 20 with __biz, mid, idx, sn
Charge$0.008$0.008

The two calls reached different accounts. The 14 accounts in the default order are mostly media outlets, such as New Weekly (3 posts), 36Kr and Huxiu (2 each), The Paper and TMTPost. The 19 accounts in the LATEST order include local tourism offices, a city management office, a campsite's own account, companies and personal travel accounts. To find new accounts in your field, use LATEST. To find the most-shared old posts on a topic, the default order fits better, but it is slow, so use async mode.

Five calls returned HTTP 503 between 12:40 and 12:47 SGT, and none was charged. Four were for the tech media account 36Kr. At 12:40:11 and 12:44:02 we asked for its past articles by WeChat ID wow36kr.

At 12:44:21 we searched for the account by name. At 12:47:19 we asked for its past articles by name. The catalog lists a WeChat ID, a gh_ ID or a link as valid input. A name is not on that list.

The fifth failure, at 12:47:01, read the body of another account's camping post through a temporary search link.

We never got any 36Kr data in this test, and we did not retry 36Kr with its gh_ ID. The past-articles call that worked at 12:56:13 was for a different account, Dingxiang Life Research, by its gh_ ID. So we have not shown that a different input format gets 36Kr through.

The 503 reply body tells your code how to retry:

FieldValue in our testWhat your code does
retryabletrueRetry only when it is true
retry_after20Wait 20 seconds, then retry
max_retries2Retry at most 2 times
recent_failures1, 2, 3 for past articlesFailures in 10 minutes
next_stepsThird failure: the fault is on the EveryInfra side if more targets failAsk the user before a ticket
supportThird failure only: ticket endpoint and request_idUse both for the ticket
billing.chargedfalseNo charge

If the retries also fail, put the target in a queue and try it again later. Do not send it again and again in the same minute. When recent_failures reaches 3, next_steps tells you to ask the user. Then pause the batch for that capability. For an account like 36Kr, you can try the other formats in the catalog: the gh_ ID or an article link. We have no evidence that they work.

The test key allows 10 calls a minute and is shared by seven writers, so two 429 replies were retried after retry_after. Treat 429 and 503 apart. The 429 reply has fault account and reason key_limit, with retry_after 9 and max_retries 3: slow down, or ask the user to raise the limit. The 503 reply has fault service: retry as in the table, and ask the user on the third failure. The six charged calls cost $0.041876 in total.

A daily routine: from search to body

Here is a routine that can run every day. Each step maps to one capability and one fixed price. We built it from the live calls above and the catalog. We did not run it end to end, and step two failed in our test. The table uses the 50-account example from the cost chart above, so the monthly numbers add up to the same $51.79.

StepCapabilityWhen it runsPrice per callMonthly cost in the example
1. Find new posts on 10 keywordsarticle_search with LATESTEvery morning$0.008$2.40
2. Turn a new account name into a gh_ IDaccount_searchOnly for a new account$0.016Not in the $51.79; $0.016 per new account
3. List new posts of each accountaccount_articlesOnce a day per account$0.008$12.00
4a. Read the six metrics of each new postarticle_metricsOnce per post$0.005556$16.67
4b. Read the bodyarticle_detailOnly when you need full text$0.005556$16.67
4c. Read the commentsarticle_commentsFor 20 chosen posts a day$0.006764$4.06

Step one gives you the titles, summaries and account names of recent posts. In our 12:56 SGT search, the 20 rows came from November 18, 2026 to November 22, 2026, and from 19 different accounts. This step finds new competitors.

Step two is the most expensive capability, but you pay for it once per account. Store the ID in your own table, so you never look it up again.

Our account search for 36Kr at 12:44:21 returned 503, so this step did not work in our test. In step three, use the permanent link to drop posts you have already seen. The LATEST links from step one already have __biz, mid, idx and sn. By the catalog rule, you can merge them with step three before you remove duplicates. We checked only the link form. We did not read a body with these links.

Step three grows with the number of accounts, and step four grows with the number of posts. In the example, steps three and four make up about 95% of the bill. To cut cost, cut bodies first. Titles and metrics already show which posts deserve a closer look. If you read bodies only for the top 10% of posts by reads, the body item falls from $16.67 to about $1.67 a month.

When another route fits better

CaseBetter routeReason
Dashboard for your own verified accountOfficial APITraffic sources by channel, openid, drafts and publishing; no call fees, only the verification fee
A few dozen bodies, no metricsSingle-post open-source tool, or copy links by handSmall volume; no paid account needed
Team already pays one TikHub bill for other platformsTikHubOne vendor; its v2 series has related articles, ad slots and account services

For your own account, no third-party API can match the official one, because the owner-only data never reaches a public page. For bodies alone, tools like jackwener/wechat-article-to-markdown turn one link into Markdown. A 2022 Juejin tutorial shows how to copy post links from the "recent posts" list of your own account.

For TikHub, the v2 Official Account series covers related articles, article ad slots and account services, which the EveryInfra catalog does not have. For WeChat Channels, it has live history and collections. EveryInfra Channels capabilities are five today: search, video, user posts, profile and comments. Some TikHub endpoints get daily volume discounts, but these Official Account entries do not. We did not call its Official Account endpoints. Before you switch, test its reply format on another endpoint with the free credit.

Law and limits

A pay-per-call API reads posts and comments that are public. That does not make every use acceptable. Three Chinese rules matter most for this data.

RuleWhat it says that matters herePractical step
Personal Information Protection LawArticle 13 allows reasonable processing of information made public lawfully; Article 27 says to stop when the person clearly refuses, and to get consent when the effect on the person is majorFor statistics, drop nicknames and avatars; keep text, like counts and provinces
Network Data Security Management RegulationsIn force from January 1, 2025, with detailed duties for data handlersKeep a record of what you collect and why
Anti-Unfair Competition Law, 2025 revisionA clause on obtaining or using other businesses' data by improper technical meansDo not rebuild a competitor's product from its data

Copying other accounts' posts to your own account is a copyright question, no matter which tool you use. Ask your own lawyer about your use case.

FAQ

What WeChat Official Account data APIs exist?

There are three kinds. The official server API (publishing, statistics, comments) works only on your own account. Open-source scrapers on GitHub use the back office search, WeChat Read, Sogou or a WeChat desktop session. Pay-per-call data APIs include EveryInfra and TikHub. To read other accounts, choose from the last two. Source of the question: Google suggestion for "Official Account data API" in Chinese (read November 22, 2026).

Which WeChat Official Account data scraping APIs do you recommend?

It depends on the job. For reports on your own account, use the official statistics API. To save a few dozen bodies, use a single-post tool such as jackwener/wechat-article-to-markdown. To read lists, read counts and comments of competitor accounts in bulk, use a pay-per-call API. EveryInfra costs $5.56 to $16.00 per 1,000 calls, and most TikHub Official Account entries cost $0.01 per call. Source: the Google suggestion "Official Account API endpoints" in Chinese (read November 22, 2026).

Which WeChat Official Account API is best?

No route covers everything. The official API gives the most for your own account and nothing for other accounts. Open source is free, but of the top two projects by stars, one stopped and one is archived. Pay-per-call removes the login and the upkeep, and you pay per call. With our numbers, watching 50 competitor accounts costs about $51.79 a month. Source: Google suggestion "WeChat Official Account API endpoints" in Chinese (read November 22, 2026).

Is there a ready-made Official Account API, so I do not have to write a scraper?

Yes. A pay-per-call API returns JSON for one HTTP request. For example, wechat_oa/article_metrics returns six numbers (reads, likes, "Wow", shares, saves, comments) for $0.005556 per call, and failures and empty results are free. You can also add the remote MCP server to Claude or Cursor and let the AI call it. Source: a V2EX thread from July 2026 that asks how people get Official Account posts with an AI agent.

Can the WeChat Official Account API read other accounts' posts?

No. The official publish, statistics and comment endpoints accept only your own account's access_token, and they return only your own content. A third-party platform also needs the other account to authorize it. Source: Google suggestions "WeChat Official Account API" and "Official Account get access token" in Chinese (read November 22, 2026).

Do the WeChat scrapers on GitHub still work?

Some work, with upkeep. The most-starred one, wechat-article-exporter, stopped on July 30, 2026, and wewe-rss is archived. The we-mp-rss project is active, but it needs a QR authorization. Single-post body tools work without a login. Source: Google suggestion "WeChat Official Account scraper GitHub" in Chinese (read November 22, 2026).

Can I get read counts and follower counts?

Read counts, yes. The pay-per-call metrics capability returns read_count. Above 100,000 it shows 100001, which matches "100k+" on the page. Follower counts are private data in the owner's back office. They are not on public pages, and the EveryInfra catalog has no such field. Source: a viewer question under YouTube video ATwEruPuAWI (read November 22, 2026, 12:50 SGT).

How do I set up an MCP server for WeChat Official Account data?

There are two ways. For your own account, use a community MCP server for account management, such as xwang152-jack/wechat-official-account-mcp, with your own AppID and AppSecret. For other accounts, add the EveryInfra remote MCP server (https://api.everyinfra.com/mcp, key in an environment variable) to your AI tool, and let the model call everyinfra_call_api. Source: Google suggestion "WeChat Official Account scraper MCP" in Chinese (read November 22, 2026).

How to start

  1. Read the free catalog: GET https://api.everyinfra.com/api/v1/social/catalog. It needs no key. Find the 13 entries with platform set to wechat_oa, and check available, price_usd and the parameters. The Official Account pages are at WeChat Official Account data API, and Channels at WeChat Channels data API.
  2. Create a key in the console. Use the curl example in the quickstart, and change the platform to wechat_oa. See prices on the pricing page. To try without code, use the playground on the market page.
  3. Start with one account. Call wechat_oa/account_articles with its gh_ ID to get permanent links. Then call wechat_oa/article_metrics and wechat_oa/article_comments for a few posts.
  4. For keyword monitoring, always pass "category":"LATEST" to wechat_oa/article_search. Raise limit as far as you need, because billing is per call.
  5. Use it in an AI tool: add the MCP config above. To compare prices with TikHub line by line, read TikHub alternatives and TikHub free credits explained.

References

  1. List published posts, WeChat Open Docs, last edited Nov 11, 2026, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/api/public/api_freepublish_batchget
  2. Publishing capability, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/guide/product/publish.html
  3. Get a published post, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/api/public/api_freepublishgetarticle
  4. Statistics API guide, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/guide/product/analysis_data/analysis_data.html
  5. Daily article summary, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/api/wedata/news/api_getarticlesummary.html
  6. Daily read data, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/api/wedata/news/api_getarticleread
  7. List comments, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/api/leaving/api_listcomment.html
  8. API overview, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/guide/dev/api/
  9. Get access token, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/api/base/api_getaccesstoken
  10. API limits, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/subscription/guide/dev/api/limit
  11. Service Account documents, WeChat Open Docs, read Nov 22, 2026, https://developers.weixin.qq.com/doc/service/guide/
  12. Pinned reply on the published-post list, WeChat developer forum, Jun 27, 2024, https://developers.weixin.qq.com/community/minihome/doc/000684561e06307dd4b10a7386fc00
  13. wechat-article/wechat-article-exporter, GitHub, read Nov 22, 2026, https://github.com/wechat-article/wechat-article-exporter
  14. wechat-article-exporter issue #200 (Jul 30, 2026), GitHub, https://github.com/wechat-article/wechat-article-exporter/issues/200
  15. cooderl/wewe-rss, GitHub, read Nov 22, 2026, https://github.com/cooderl/wewe-rss
  16. chyroc/WechatSogou, GitHub, read Nov 22, 2026, https://github.com/chyroc/WechatSogou
  17. rachelos/we-mp-rss, GitHub, read Nov 22, 2026, https://github.com/rachelos/we-mp-rss
  18. we-mp-rss issue #485 (Oct 30, 2026), GitHub, https://github.com/rachelos/we-mp-rss/issues/485
  19. we-mp-rss issue #489 (Nov 19, 2026), GitHub, https://github.com/rachelos/we-mp-rss/issues/489
  20. wnma3mz/wechat_articles_spider, GitHub, read Nov 22, 2026, https://github.com/wnma3mz/wechat_articles_spider
  21. bowenpay/wechat-spider, GitHub, read Nov 22, 2026, https://github.com/bowenpay/wechat-spider
  22. jackwener/wechat-article-to-markdown, GitHub, read Nov 22, 2026, https://github.com/jackwener/wechat-article-to-markdown
  23. yeximm/Access_wechat_article, GitHub, read Nov 22, 2026, https://github.com/yeximm/Access_wechat_article
  24. wonderfulsuccess/weixin_crawler, GitHub, read Nov 22, 2026, https://github.com/wonderfulsuccess/weixin_crawler
  25. Moore-developers/moore-wechat-article-downloader, GitHub, read Nov 22, 2026, https://github.com/Moore-developers/moore-wechat-article-downloader
  26. xwang152-jack/wechat-official-account-mcp, GitHub, read Nov 22, 2026, https://github.com/xwang152-jack/wechat-official-account-mcp
  27. Sogou WeChat search, read Nov 22, 2026, https://weixin.sogou.com/
  28. TikHub public price list, read Nov 22, 2026, 12:47 SGT, https://api.tikhub.io/api/v1/tikhub/user/get_all_endpoints_info
  29. TikHub FAQ, read Nov 22, 2026, https://tikhub.io/faq
  30. TikHub MCP, read Nov 22, 2026, https://tikhub.io/mcp
  31. TikHub docs, read Nov 22, 2026, https://docs.tikhub.io/
  32. EveryInfra social data catalog, read Nov 22, 2026, 12:37 SGT, https://api.everyinfra.com/api/v1/social/catalog
  33. Model Context Protocol, https://modelcontextprotocol.io/
  34. Personal Information Protection Law of the PRC, full text reposted by the National Bureau of Statistics, https://www.stats.gov.cn/gk/tjfg/xgfxfg/202503/t20250310_1958923.html
  35. Network Data Security Management Regulations, Cyberspace Administration of China, Sep 30, 2024, https://www.cac.gov.cn/2024-09/30/c_1729384452307680.htm
  36. Anti-Unfair Competition Law of the PRC (2025 revision), China National Intellectual Property Administration, https://www.cnipa.gov.cn/art/2026/5/20/art_104_206437.html
  37. How do you get WeChat Official Account posts with an AI agent? V2EX, Jul 2026, https://www.v2ex.com/t/1198827
  38. I tried six ways to get WeChat Official Account posts, Juejin, Apr 2026, https://juejin.cn/post/7624018983559397376
  39. How to get the history links of your Official Account, Juejin, 2022, https://juejin.cn/post/7106768099361161223
  40. Monitor Official Accounts in real time with an AI workflow, YouTube, Jul 3, 2025, https://www.youtube.com/watch?v=ATwEruPuAWI
  41. WeChat Official Accounts Platform, read Nov 22, 2026, https://mp.weixin.qq.com/

Reference BL-211

Google Analytics needs consent. Privacy