Bilibili Scraper API: Videos, Comments, Danmaku & UP主 (B站 哔哩哔哩)

Bilibili Scraper API: Videos, Comments, Danmaku & UP主 (B站)

Get Bilibili (哔哩哔哩, B站) videos, comments, danmaku summaries, and creator profiles without a login, cookie, or API key. Rows carry title, stats, comment text, or a danmaku summary for creator research and comment analysis. Risk-control blocks return an error row and can leave the run partial; earlier rows are kept.

• Low per-row price: $2 per 1,000 rows, and you only pay for rows you get.
• Optional AI brand/spam labels: tag comments and videos with brand and spam labels for faster social listening.
• Change-only monitoring: schedule it daily and get only new or changed videos, comments and creators.
• Reliable: tested every day so scheduled runs keep working.

中文简介：B站爬虫 / 哔哩哔哩数据采集 / B站API替代：B站数据采集，无需登录、无需 Cookie。支持关键词搜索视频、视频详情（播放、点赞、投币、收藏、分享、弹幕数）、评论与楼中楼回复、热门视频、UP主资料（粉丝数、投稿数、获赞数）和弹幕分析（弹幕密度、高峰分钟、弹幕样本）。按条计费，每 1000 条 2 美元。

For AI agents

Call this actor through the Apify MCP server. maxChargeUsd is this actor's cap. MCP callOptions.maxTotalChargeUsd is a separate platform cap; set both before a paid call. Do not use mode: "selftest" — it writes a maintenance record, not video rows.

List prices (unchanged): run start $0.001, each video, comment, reply, or creator $0.002, each danmaku summary $0.01. changesOnly adds $0.002 per new or changed row. AI labels add $0.0005 per labeled video. Five search rows are about $0.011. A risk-control error row is free.

outputFormat: "compact" also writes key OUTPUT_COMPACT (columnar-v1: columns, missing, rows). A cell equal to missing was absent. The dataset is the row fallback. Same rows, same price.

Search:

{ "mode": "search", "searchQueries": ["原神"], "maxResults": 5, "sortOrder": "totalrank", "maxChargeUsd": 1 }

Fields: type video, bvid, title, viewCount, likeCount, danmakuCount, replyCount, authorName, publishDate, url.

One video:

{ "mode": "video_detail", "videoUrls": ["BV1xx411c7mD"], "maxChargeUsd": 1 }

Adds coinCount, favoriteCount, shareCount, cid, parts, hasSubtitles.

Comments:

{ "mode": "video_comments", "videoUrls": ["BV1xx411c7mD"], "maxComments": 5, "includeReplies": false, "maxChargeUsd": 1 }

Fields: type comment or reply, text, likeCount, authorName, createdAt, videoBvid.

Popular:

{ "mode": "popular", "maxResults": 5, "maxChargeUsd": 1 }

Same video fields, source popular.

Creator:

{ "mode": "user_profile", "userIds": ["546195"], "maxChargeUsd": 1 }

Fields: type user, name, followers, videoCount, totalLikes, level, isVerified, profileUrl.

Danmaku summary (one row, $0.01, not one row per comment):

{ "mode": "video_detail", "videoUrls": ["BV1xx411c7mD"], "includeDanmaku": true, "danmakuSampleSize": 5, "maxChargeUsd": 1 }

Fields: type danmaku, danmakuFetched, danmakuPerMinute, danmakuPeakMinute, danmakuSample.

Try it in one click

The input form on this page is already prefilled with the example below — press Start and it runs as-is.

{ "mode": "search", "searchQueries": ["原神"], "maxResults": 10, "sortOrder": "totalrank", "maxChargeUsd": 0.5 }

What you get:

• Up to 10 "原神" (Genshin Impact) videos ranked by relevance: views, likes, coins, favorites, shares, danmaku count, author, publish date.
• About 10 rows for roughly $0.02 at the listed pricing (run-start $0.001 + $0.002/row).
• Add includeComments or switch to video_comments mode for comment rows at the same per-row price.

Why Bilibili data matters

• Scale and attention: 371 million monthly and 116.5 million daily active users, spending an average 113 minutes a day on the platform (Q2 2026, Bilibili results).
• Who's there: the average user is 26.5 years old, and the company describes them as China's main consumption force; its top ad verticals are games, consumer electronics, internet services, e-commerce and automotive (Q2 2026 earnings call).
• What it's used for: long-form reviews, tutorials, gaming, anime and tech. Comments and danmaku (弹幕) show second-by-second reactions you can't get anywhere else.
• For brands: find the creators (UP主) who move your category, and read what Gen Z actually says about your product under review videos.
• For market researchers and AI teams: engagement benchmarks, creator discovery, and a deep corpus of short, opinionated Chinese comments for sentiment and LLM pipelines.

What Bilibili data you can scrape

Mode	What you get	Input
search	Videos matching keywords (sort, length and date filters)	searchQueries
video_detail	Full stats: views, likes, coins, favorites, shares, danmaku, comments, parts, hasSubtitles	videoUrls (links or BV ids)
video_comments	Top-level comments (popular or newest first) plus optional replies	videoUrls
popular	Bilibili's popular / trending board	(none)
user_profile	Creator (UP主) followers, following, video count, total likes, level, verification, bio	userIds (mid or space.bilibili.com link)

Add-ons for search, video_detail and popular: includeComments, includeReplies, includeDanmaku, includeVideoDetails.

Pricing (pay per event)

Event	Price
Run start	$0.001 per run
Row (video, comment, reply or creator)	$0.002 ($2 per 1,000)
Danmaku summary (one row per video, only with includeDanmaku)	$0.01

Error rows are free. Set maxChargeUsd (default $5) and the run stops cleanly before it goes over; rows already delivered are kept.

Example costs: 100 search results = $0.20. 10 videos × 50 comments = 510 rows ≈ $1.02. Danmaku summaries for 50 videos = $0.50.

Works on Apify’s Free plan — $5/mo platform usage, no credit card required to try. Per-row prices above are unchanged.

Input examples

Search, newest first, with comments:

{
  "mode": "search",
  "searchQueries": ["机械键盘", "原神"],
  "maxResults": 50,
  "sortOrder": "pubdate",
  "publishedAfter": "2026-10-01",
  "includeComments": true,
  "maxComments": 20
}

Comments and replies of specific videos:

{
  "mode": "video_comments",
  "videoUrls": ["https://www.bilibili.com/video/BV1xx411c7mD"],
  "maxComments": 200,
  "sortComments": "hot",
  "includeReplies": true,
  "maxRepliesPerComment": 10
}

Video details with a danmaku summary:

{ "mode": "video_detail", "videoUrls": ["BV1xx411c7mD"], "includeDanmaku": true }

Creator profiles (schedule it daily to track follower growth):

{ "mode": "user_profile", "userIds": ["452412746", "https://space.bilibili.com/265504053"] }

Output examples

Video:

{
  "type": "video", "source": "detail", "bvid": "BV1frJB64EQv", "aid": 117456668459441, "cid": 42824894014,
  "title": "2026年双十一高性价比机械键盘选购推荐", "url": "https://www.bilibili.com/video/BV1frJB64EQv",
  "thumbnailUrl": "https://i1.hdslb.com/bfs/archive/....jpg", "duration": 2295, "parts": 1,
  "viewCount": 2179, "likeCount": 106, "coinCount": 8, "favoriteCount": 151, "shareCount": 12,
  "danmakuCount": 98, "replyCount": 41, "authorName": "UP主名", "authorMid": 12345678,
  "category": "数码", "tags": ["机械键盘"], "publishDate": "2026-10-16T10:00:00+00:00",
  "hasSubtitles": true, "scrapedAt": "2026-10-18T15:48:14+00:00"
}

Comment or reply:

{
  "type": "reply", "videoBvid": "BV1frJB64EQv", "commentId": 321073847136, "rootCommentId": 321056652768,
  "text": "售后换呗", "likeCount": 1, "replyCount": 0, "createdAt": "2026-10-18T02:47:21+00:00",
  "authorName": "景行推荐", "authorMid": 2133808831, "scrapedAt": "2026-10-18T15:48:14+00:00"
}

Danmaku summary (real row from a test run on 2026-10-18):

{
  "type": "danmaku", "videoBvid": "BV1qm41197pQ", "danmakuFetched": 1101, "danmakuTotal": 1303,
  "danmakuPoolCapped": true, "danmakuUniqueSenders": 875, "danmakuPerMinute": 87.96,
  "danmakuPeakMinute": 4, "danmakuPeakMinuteCount": 183, "danmakuFirstAt": 4.8, "danmakuLastAt": 748.3,
  "danmakuSample": [{"atSecond": 4.8, "text": "不是还有个茶轴吗？"}, {"atSecond": 36.8, "text": "惊人收藏"}]
}

Creator (real row, 2026-10-18):

{
  "type": "user", "mid": 452412746, "name": "Yooupi食途", "profileUrl": "https://space.bilibili.com/452412746",
  "bio": "一个上山打虎，下海捉蛟的清奇频道。…", "level": 6, "followers": 2323150, "following": 126,
  "videoCount": 278, "totalLikes": 15814347, "isVerified": true, "verificationTitle": "2023百大UP主、知名UP主"
}

Compact output for AI agents

Set Output format to compact and the run still pushes the same rows to the dataset at the same price — nothing about billing changes. It additionally writes OUTPUT_COMPACT to the run's key-value store: our token-optimized, lossless encoding of the same rows, built to be cheaper to hand to an LLM than the plain dataset JSON.

Measured on a real 406-row export of this actor's comment/reply rows:

Tokenizer	Pretty JSON	Minified JSON	Compact	vs Pretty	vs Minified
cl100k_base (GPT-4/3.5)	74,310	59,673	45,538	38.7% smaller	23.7% smaller
o200k_base (GPT-4o)	67,672	53,034	39,320	41.9% smaller	25.9% smaller

Need regular row objects in your own code? The normal dataset is always there, unchanged, alongside the compact copy.

Monitor mode

Schedule this actor and set Output mode to changesOnly to get a diff instead of a full re-export: the first run records a baseline, and every run after that reports only videos, comments, creators, or danmaku summaries that are new, changed (with a changedFields list), or removed since last time. Removed rows carry just the item's id. Use Monitor name if you run more than one monitor and want to keep their baselines separate, and Report removed items / Fields to ignore to tune what counts as a change. Default mode (all) is unaffected — it always exports everything, every run. In changesOnly, every new/changed row (including danmaku summaries) is billed as a changed row in addition to its normal per-row charge — see Pricing below; a removed row is billed as a changed row only (there's no per-row charge for an item that wasn't actually fetched this run); unchanged rows are free and aren't pushed.

AI labels (optional)

Turn on Add AI labels to add AI brand-relevance and spam filtering to each video row (search, video detail, and popular — not comments, danmaku, or profiles): a brandRelevant flag (only when Brand to match is set) and a spam flag, each with its own confidence score. Below a label's confidence cutoff, that label comes back as low_confidence instead of a guess — never a coin-flip label. This is billed per video actually labeled, in addition to that row's own charge; it never blocks a row's own delivery if the run's budget is tight — see Pricing below.

Good to know

• Public data only. The actor never logs in. Comment rows carry only the public nickname and mid; no avatars or other personal fields.
• Subtitles: Bilibili only serves subtitle and AI-subtitle text to logged-in users, so this actor reports hasSubtitles (whether tracks exist) and does not download the text.
• Creator upload lists (space pages) aren't available without login. To find a creator's videos, use search with their name.
• Search depth: Bilibili caps one keyword at about 1,000 results. Use date windows or more specific keywords to go further.
• Danmaku: Bilibili keeps a capped pool per video. danmakuPoolCapped tells you when the summary covers part of the total.
• Reliability: a self-test runs every day against live Bilibili, so breakages get spotted and fixed fast. If Bilibili blocks every request in a run, the run is marked failed instead of ending green with 0 rows, so your schedule alerts fire.

Run it on a schedule

Suggested frequency: daily for the trending board and keyword tracking, weekly for creator (UP主) stats. Monitor mode returns only new or changed rows.

To set it up: save your input as a Task in Apify Console, then add a Schedule. Send results where you work with Apify integrations: Google Sheets, Slack or email notifications, webhooks, Make / Zapier / n8n, or straight into an AI agent through the Apify MCP server. A run that can't get data is marked failed, so your alerts fire.

FAQ

Is there an official Bilibili API? Not for international developers. This actor gives you the same public data any visitor sees, as clean structured rows.

English or Chinese keywords? Bilibili indexes Chinese text. Searching 耐克 returns many more videos than "Nike".

How do I schedule it? Save your input as a Task in Apify Console, then add a Schedule (for example 0 8 * * * for daily at 8 AM).

Feedback

If the data helped, a short rating on the Reviews tab helps others find this actor. Something broke or missing a field? Open an issue on the Issues tab — we aim to reply within 24 hours.

Related actors

• Weibo Scraper — posts, comments and hot search from China's biggest microblog.
• Douban Scraper — movie and book ratings, comments and reviews.
• WeChat Articles Scraper — WeChat Official Account articles with full text.
• AliExpress Scraper — AliExpress products, prices and variants.
• YouTube Scraper — YouTube videos and comments, for comparing audiences outside China.
• TikTok Scraper — TikTok videos and creators.

关键词： B站爬虫, 哔哩哔哩采集, B站评论采集, 弹幕分析, B站热门, UP主数据, Bilibili API, Bilibili comments scraper, danmaku scraper.