📺 Bilibili Scraper (哔哩哔哩 / B站): Videos, Comments, Creators and Rankings

Returns Bilibili videos with views, likes, coins, favourites, tags and creator, the comments and replies under a video, a creator's complete upload list and profile, the trending board, the top 100 of every category, and danmaku (弹幕). Search by keyword in Chinese or English, give creators or video links, or pick a ranking. It needs no login, no cookies and no browser. Export to JSON, CSV or Excel, run on a schedule, call via API, or connect to Make, Zapier or n8n.

Price: $4.00 per 1,000 videos or creator profiles and $1.00 per 1,000 comments on the Apify Free plan, down to $3.40 and $0.85 from Gold up. Not charged: a search without results, a creator or video link that does not exist, a video's comments when they are closed.

What other Bilibili scrapers state as their limits, and what this one does

Their documentation says	This Actor
A creator's videos are found by searching the channel name, because Bilibili's upload list refuses requests	Reads the upload list itself, from the newest video to the first: 1,442 of 1,442 and 678 of 678 in the tests
Deep comment paging needs a logged-in session; newest-first returns about three comments	Pages the whole comment list in hot or newest order: 1,392 comments and replies of a video Bilibili counts 1,390 on, 5,000 newest comments of another in 80 seconds
The category is empty for a video given by link	category is filled for video links as well
Coins and shares are missing in search and creator lists	includeDetails adds them, with tags and the creator's followers, at no extra charge
Category rankings answer an error, so a category is picked out of the trending board	The top 100 of each of 18 categories and the overall top 100, as Bilibili ranks them
One price for every row, comments included: $0.002 to $0.02, depending on the Actor and the plan	$0.004 a video, $0.001 a comment

One Bilibili search lists about 1,000 videos. Sorted by newest, this Actor continues behind the oldest video read: 2,300 videos of one keyword in 24 seconds.

🔎 What is the Bilibili Scraper?

Bilibili is China's largest video community: uploads by creators (UP主), a comment section under every video and danmaku, the comments that scroll across a video at a chosen second. The site offers no public API to developers outside China. This Actor calls the same data requests that Bilibili's website and phone app make and returns clean rows.

What data does it extract?

• Videos: title, description, cover, length, views, likes, coins, favourites, shares, number of comments and danmaku, publish time (UTC), category, tags, parts, place of publication
• Creators: name, signature, followers, following, total likes, number of videos, level, verification text, live room
• Comments and replies: text, likes, number of replies, time, author name, id, level and avatar, and for a reply the comment it stands under
• Danmaku: each one's second in the video, text, colour and time of sending
• Rankings: the trending board (about 600 videos) and the top 100 of each category

⬇️ Input

Fill in one or several of the first four fields.

Field	Type	Default	What it does
queries	array		Search keywords, Chinese or English
creators	array		Space links, UIDs or exact names of creators
videoUrls	array		Video links, BV ids, av numbers, b23.tv short links
categories	array		popular (trending board), all (top 100 overall), or animation, music, dance, game, knowledge, tech, sports, car, life, food, animal, kichiku, fashion, entertainment, film, movie, tv, guochuang
maxResults	integer	20	Videos per search, creator or ranking, up to 20,000
sortBy	string	relevance	relevance, newest, views, danmaku, favorites, comments; for creators views or newest
duration	string	any	under10, 10to30, 30to60, over60 (minutes), for searches
publishedAfter, publishedBefore	string		A day such as 2026-09-01 (UTC) or an age such as 7 days; for searches and creators
creatorOutput	string	videos	videos, profile (one row per creator, no videos) or both
includeDetails	boolean	false	Full record for every video of a search, creator or ranking
maxCommentsPerVideo	integer	0	Comments to save under each video, up to 20,000
commentsSort	string	hot	hot or newest
includeReplies	boolean	false	Replies under the comments, as rows of their own
includeDanmaku	boolean	false	One row per video with its danmaku
onlyNew, monitorId	boolean, string	false	Skip videos an earlier run of the same monitor delivered

Example input

Videos for a keyword, newest first, from the last week:

{
  "queries": ["黑神话悟空"],
  "sortBy": "newest",
  "publishedAfter": "7 days",
  "maxResults": 500
}

Every upload of two creators, with their profiles:

{
  "creators": ["https://space.bilibili.com/546195", "9617619"],
  "creatorOutput": "both",
  "maxResults": 2000
}

A video with its newest 1,000 comments and their replies:

{
  "videoUrls": ["https://www.bilibili.com/video/BV1feaH63Epk"],
  "maxCommentsPerVideo": 1000,
  "commentsSort": "newest",
  "includeReplies": true
}

The 100 top videos in tech and in games, with full records:

{
  "categories": ["tech", "game"],
  "maxResults": 100,
  "includeDetails": true
}

⬆️ Output

Every row has a type: video, comment, user or danmaku. The dataset has a view for videos, one for comments and one for creators. All samples are from runs of October 5, 2026.

A video (here with its full record):

{
  "type": "video",
  "bvid": "BV1feaH63Epk",
  "aid": 117354226849930,
  "url": "https://www.bilibili.com/video/BV1feaH63Epk",
  "title": "我在一天内学会了5个乐器！！",
  "description": "别管难听不难听，你就说发没发出声来吧。",
  "thumbnailUrl": "https://i2.hdslb.com/bfs/archive/462781c040fdb9dd4584f0ba9cf4c00cd1fba795.jpg",
  "duration": 727,
  "durationFormatted": "12:07",
  "viewCount": 2468667,
  "likeCount": 174038,
  "coinCount": 15272,
  "favoriteCount": 24917,
  "shareCount": 1777,
  "danmakuCount": 3438,
  "replyCount": 1389,
  "authorName": "老番茄",
  "authorMid": 546195,
  "authorFollowers": 20821588,
  "publishDate": "2026-09-30T03:30:00+00:00",
  "publishTimestamp": 1790739000,
  "category": "单机游戏",
  "tags": ["搞笑", "乐器", "萨克斯", "钢琴", "架子鼓", "老番茄"],
  "parts": 1,
  "cid": 42310699872,
  "publishLocation": "上海",
  "source": "videos:video links",
  "rank": 1,
  "scrapedAt": "2026-10-05T09:16:40.646953+00:00"
}

Which fields a list fills without includeDetails: a search has everything except coins, shares, the creator's followers, parts, cid and publishLocation; a creator's upload list has title, cover, length, views, danmaku count, publish time, category and the creator; a category ranking has likes, favourites and comments as well; the trending board has the full record except tags. A field a list does not carry is null, never 0.

A comment (author replaced here; a reply carries isThreadReply: true and the id of the comment it stands under):

{
  "type": "comment",
  "commentId": 315527976641,
  "text": "…",
  "likeCount": 51,
  "replyCount": 0,
  "createdAt": "2026-09-30T07:17:17+00:00",
  "createdTimestamp": 1790752637,
  "authorName": "example_user",
  "authorMid": 12345678,
  "authorAvatar": "https://i0.hdslb.com/bfs/face/example.jpg",
  "authorLevel": 6,
  "videoBvid": "BV1feaH63Epk",
  "videoAid": 117354226849930,
  "isThreadReply": true,
  "rootRpid": 319089854560,
  "parentRpid": 319089854560,
  "scrapedAt": "2026-10-05T09:16:40.227311+00:00"
}

A creator:

{
  "type": "user",
  "mid": 546195,
  "name": "老番茄",
  "face": "https://i0.hdslb.com/bfs/face/bc5ca101313d4db223c395d64779e76eb3482d60.jpg",
  "sign": "新浪微博：_老番茄_",
  "level": 6,
  "fans": 20821588,
  "friend": 6,
  "attention": 6,
  "isOfficial": true,
  "officialDesc": "2025百大UP主、2025年度商业影响力奖UP主、2022年度联合创作奖UP主",
  "archiveCount": 678,
  "likeCount": 195572013,
  "profileUrl": "https://space.bilibili.com/546195",
  "isLive": true,
  "liveRoomUrl": "https://live.bilibili.com/21852",
  "source": "creator:546195",
  "scrapedAt": "2026-10-05T09:16:38.344466+00:00"
}

A video's danmaku (two of 1,800 shown):

{
  "type": "danmaku",
  "videoBvid": "BV1feaH63Epk",
  "videoAid": 117354226849930,
  "cid": 42310699872,
  "danmakuCount": 1800,
  "danmakuOnVideo": 3438,
  "danmaku": [
    {"time": 0.0, "text": "《看了场交响乐 啥乐器都想学》", "sentAt": "2026-09-30T05:03:56+00:00", "color": "#ffffff"},
    {"time": 0.0, "text": "《三分钟热度》", "sentAt": "2026-10-01T07:48:26+00:00", "color": "#ffffff"}
  ],
  "scrapedAt": "2026-10-05T09:16:40.632601+00:00"
}

A video's row is saved after its comments and its danmaku row. The run's SUMMARY record lists, per search, creator and ranking, how many videos were saved, how many Bilibili has, and how the reading ended.

💰 What a run costs

Apify plan	Per 1,000 videos, creator profiles or danmaku lists	Per 1,000 comments or replies
Free	$4.00	$1.00
Bronze	$3.80	$0.95
Silver	$3.60	$0.90
Gold, Platinum, Diamond	$3.40	$0.85

A video row costs two events, the row ($0.001) and the record ($0.003); a comment row costs the row only. Apify's standard start event ($0.00005 per run at the default memory) is the only other charge. includeDetails costs nothing extra. With a maximum charge set for the run, the Actor stops when the next row would pass it, and a video that was started still gets its own row.

Examples on the Free plan: 500 videos for a keyword, $2.00. The 678 uploads of one creator with the profile, $2.72. One video with 5,000 comments, $5.00. A daily check of 50 creators that finds 30 new videos, $0.12.

⏱️ Measured (October 5, 2026)

Run	Rows	Time
Sample input: 20 videos for one keyword	20	3 s
Five keywords, 300 videos each	1,500	13 s
One keyword, newest first, past Bilibili's list limit	2,300	24 s
One keyword, most views, September 2026 only (Bilibili's limit for one search)	1,050	14 s
Every upload of one creator (1,442 of 1,442)	1,442	24 s
Ten creator profiles	10	3 s
Top 100 of one category with full records	100	32 s
One video with every comment and reply (Bilibili counts 1,390)	1,393	25 s
One video with its 5,000 newest comments	5,001	80 s

Measured from a server in Germany without a proxy, at 85 MB of memory in a 3,500-row run. A run killed after 7 seconds in the middle of ten videos with 200 comments each, and started again on the same dataset, ended with 10 videos and 2,000 comments, none of them twice.

🔔 Monitoring: only what is new

Turn on onlyNew, give the run a monitorId, and put it on a schedule. Every video a run saves is remembered under that name; later runs skip it. With sortBy: newest a search stops reading two pages after the last new video, so a daily check of a keyword or of fifty creators takes a few seconds and costs only the new videos. A watch starts at its first run: videos older than the oldest one the first run saved are not delivered later. The memory is a key-value store named bilibili-monitor in your own account.

{
  "creators": ["546195", "9617619"],
  "onlyNew": true,
  "monitorId": "creators-daily",
  "maxResults": 50
}

🤖 For AI agents

Smallest useful calls:

{"queries": ["人工智能"], "maxResults": 20}

{"videoUrls": ["BV1feaH63Epk"], "maxCommentsPerVideo": 50}

Every row has type (video, comment, user, danmaku). All four target fields are arrays and can be combined in one run. No credentials are needed. A target that does not exist returns no row, and the status message and SUMMARY name it. A video is saved once per run, also when two searches find it.

💡 Use cases

• 📈 Brand and product monitoring in China: what is said about a brand, a game or a film this week, with comments and danmaku as reaction data.
• 🧑‍🎤 Creator (KOL) research: followers, likes, upload history and views per video for a list of creators, before a campaign and during it.
• 🧠 Chinese text for analysis and model training: comments, replies and danmaku by topic, with times and like counts.
• 🔥 Trend tracking: the trending board and the top 100 of each category, every day.

⚠️ Limits, stated plainly

• The times above are from a test server, not from Apify's network. The Actor paces its requests and moves to Apify's datacenter proxy, then to residential addresses, when Bilibili turns an address away.
• One search lists about 1,050 videos. Sorted by newest the Actor continues behind the oldest video read; in any other order the search ends there and the status message says so. Narrow the dates to cover more.
• Bilibili's search is loose. A query without good matches still returns videos. Check title and tags.
• includeDetails reads about five videos a second. Bilibili turns an address away for twenty minutes and more when the video record is asked for in bursts.
• Danmaku: what Bilibili's danmaku file of a video holds, which is a selection on busy videos (1,800 of 3,438 on the test video), and of a video's first part only.
• Comments: what a visitor can read. A video with comments closed returns none. The author's IP region is not given to visitors.
• Not available without login: subtitles and transcripts, a creator's followers list, favourites, watch history.
• Creator by name takes the creator Bilibili's search puts first, an exact name before the others; SUMMARY says which one was matched. A link or UID is exact.
• Times are UTC. Bilibili shows Beijing time (UTC+8).

❓ FAQ

Do I need a Bilibili account, cookies or an API key? No.

Is there an official Bilibili API? Not one open to developers outside China: Bilibili's developer platform asks for verification with a Chinese phone number. This Actor reads what the website and the app show to a visitor.

Can I use the input I wrote for another Bilibili Actor? Mostly. mode (search, video_detail, video_comments, user_videos, user_profile, popular), searchQuery, userIds, sortOrder, durationFilter, pubtimeBegin, pubtimeEnd, maxComments, sortComments, includeComments, deltaMode, deltaStateKey, keywords, searchTerms, startUrls, userUrls and maxItems are read. The output uses the same field names as the most-used Bilibili Actor (bvid, viewCount, coinCount, authorMid, publishDate, commentId, rootRpid, fans, archiveCount). Differences: a comments run also returns the video's own row, and a creator run returns the profile only when creatorOutput asks for it.

How do I get everything for a keyword? sortBy: newest with a large maxResults. Add publishedAfter to stop at a date.

Why does a video appear only once when two of my searches find it? A video is saved, and charged, once per run, under the first search that found it. source says which.

How fresh are the counts? They are read at the moment of the run; scrapedAt is on every row.

What happens at a spending limit or a timeout? The run stops cleanly, keeps its rows and says in the status message where it stopped. Started again on the same dataset (resurrected), it continues without saving a row twice.

How do I call it from code? With the Apify API or the Python and JavaScript clients; every run returns a dataset you can fetch as JSON or CSV. It also works as a tool through Apify's MCP server.

Is it legal? The Actor reads what Bilibili shows to every visitor without login. Comment and danmaku authors are people: their names and what they wrote are personal data, so GDPR, China's PIPL and similar rules apply to how you store and use them. Respect Bilibili's terms and the copyright of creators; the Actor returns data about videos, not the videos.

🔗 You might also like

• YouTube Comments Scraper: comments and replies of YouTube videos
• TikTok Comments Scraper: comments and replies of TikTok videos
• TikTok Profile Videos Scraper: every video of a TikTok profile with statistics
• Telegram Channel Scraper: posts of public Telegram channels

Keywords: bilibili scraper, bilibili api, 哔哩哔哩 爬虫, B站 数据, bilibili video scraper, bilibili comments scraper, bilibili danmaku, 弹幕, bilibili creator data, UP主 数据, bilibili search api, bilibili trending, bilibili ranking, china social media scraper, chinese video data, bilibili kol monitoring, bilibili comment export