All crawlers

Alibaba · Training

QwenBot

Alibaba confirms the activity and never names the actor. Its Model Studio disclosure states that Qwen and Wan are trained partly on "Publicly available internet data — web content accessible to the general public", and adds the unusually frank admission that datasets "may include copyrighted, trademarked, or patented content, as well as public domain content", so corpus-building from the open web is vendor-confirmed. Missing is every link between that corpus and a requesting agent: no user-agent token, no fetch policy, no crawl-rate figure, no way for a site owner to tell a training sweep apart from any other Alibaba traffic. The string that circulates for this token advertises `+https://tongyi.aliyun.com/bot` as its own reference URL, and that URL 301s to a 404: a self-documenting agent whose documentation does not exist is a reason to distrust the string, not to reprint it, which is why no example appears here.

Operated by
Alibaba
Purpose
Training

Collects pages into a corpus used to train models.

robots.txt token
QwenBot
Verification
User agent only

How to verify QwenBot

Alibaba publishes no list of IP addresses for QwenBot, so nobody (us included) can prove that a request carrying this user agent really came from Alibaba. The name is trivial to copy. Treat it as a claim the visitor is making about itself, not as an identity anyone has checked.

There is no range list to confirm. This entry was last reviewed against Alibaba's own documentation on .

User agent

Alibaba has not published a full user-agent string for QwenBot. Requests are identified by the QwenBot product token appearing in the User-Agent header; we match that token rather than a whole string, because the rest of the header varies and matching it would miss real traffic.

robots.txt for QwenBot

Block

User-agent: QwenBot
Disallow: /

Allow

User-agent: QwenBot
Allow: /

robots.txt is a request, not an enforcement mechanism. It is honoured by convention, and a crawler that ignores it is stopped at your edge, not in a text file.

What blocking QwenBot costs you

A `Disallow` costs no referral traffic by construction: a training fetch returns nobody to your site, generates no citation and produces no link. The trade is that your material stops refreshing inside future Qwen weights while already-ingested copies stay ingested, making the decision forward-only and never retroactive. The weak half is enforcement: Alibaba publishes no opt-out mechanism, no deletion path, and no statement that a robots.txt rule is honoured by whatever performs the collection, so a publisher who blocks is relying on nothing the vendor has actually promised.

Vendor documentation

Alibaba does not publish documentation for QwenBot that we could find. Everything on this page comes from what they do publish elsewhere and from observed behaviour, so treat it accordingly.

Other Alibaba tokens we track