TongyiBot
Alibaba's crawler for the Qwen model family - one of the most widely deployed open-weight model lines anywhere.
Reference reviewed by Sona
The operator publishes little or no documentation for this token. Details here come from community crawler registries and observed behavior, not vendor confirmation.
| Operator | Alibaba |
|---|---|
| Powers | Qwen and Tongyi Qianwen model content collection |
| Purpose | Model training |
| User-agent token | TongyiBot |
| Respects robots.txt | Mostly |
TongyiBot collects web content for Alibaba's Tongyi Qianwen AI, the Qwen model family. Qwen deserves more attention than its crawler's obscurity suggests: it is among the most downloaded and fine-tuned open-weight model families in the world, widely used well outside China as a base for other products.
That open-weight distribution changes what allowing this token means. When a closed model trains on your content, the exposure ends with that vendor. When an open-weight family does, the weights are downloaded, fine-tuned, and redeployed by thousands of third parties - so the reach of a single crawl is considerably wider and entirely untraceable.
Documentation does not match that significance. Alibaba publishes no crawler page, no IP feed, and no compliance policy; community registries describe TongyiBot as fetching content for the Tongyi Qianwen assistant and list robots.txt compliance as unclear.
Treat compliance as expected but unverified. Add the rule, then check your logs to see whether it is honored, and plan on server-level enforcement if the block genuinely matters to you.
How TongyiBot behaves
- No operator crawler documentation, IP feed, or compliance policy.
- Community registries record robots.txt compliance as unclear.
- Feeds the Qwen family, whose open weights are redistributed and fine-tuned by many third parties.
- Traffic generally originates from Alibaba Cloud ranges shared with unrelated workloads.
Allow TongyiBot
Your content can be represented in Qwen models, which power a very large ecosystem of applications in and beyond China.
User-agent: TongyiBot Allow: /
Block TongyiBot
Keep your content out of a model family whose open weights are redeployed by thousands of untraceable third parties.
User-agent: TongyiBot Disallow: /
How to verify TongyiBot
No published verification method
Alibaba publishes no IP range feed or reverse-DNS convention for TongyiBot, so its traffic cannot be authenticated. Requests generally originate from Alibaba Cloud ranges, which some operators treat as a rough signal - but Alibaba Cloud hosts an enormous amount of unrelated activity, so that inference is weak and will catch bystanders.
Check an IP against this botCommonly confused with TongyiBot
Another Chinese lab's training crawler with similarly thin documentation - but ChatGLM-Spider's operator is not even confirmed, while TongyiBot is clearly Alibaba's.
Huawei's equivalent: a major model line behind an undocumented crawler with no verification method.
TongyiBot FAQs
What is Qwen and why does it matter here?
Qwen is Alibaba's Tongyi Qianwen model family, among the most downloaded and fine-tuned open-weight model lines in the world. Its open distribution means content that influences it reaches far more deployments than a single vendor's closed model would.
Does TongyiBot respect robots.txt?
Unconfirmed. Alibaba publishes no compliance policy and community registries list it as unclear. Add the rule, then verify against your own logs rather than assuming it is honored.
Can I verify TongyiBot traffic?
No. There is no published IP feed or reverse-DNS convention. Requests typically come from Alibaba Cloud ranges, but those host vast amounts of unrelated traffic, so that is a weak signal rather than verification.
Why does open-weight training change the decision?
Because the exposure does not stop at Alibaba. Open weights get downloaded, fine-tuned, and redeployed by thousands of third parties, so one crawl propagates into products you will never be able to identify.
Can TongyiBot read your page right now?
Test any URL and see exactly what AI crawlers receive.