Baidu · Training
ERNIEBot
Nine rows is how many crawler user agents Baidu enumerates in its webmaster FAQ (`Baiduspider` for PC, mobile and other search, then `-image`, `-video`, `-news`, `-favo`, `-cpro` and `-ads`), and this token appears in none of them. That is stronger evidence than silence: a vendor that maintains a public user-agent table and omits a name is telling you something a vendor with no table at all cannot. ERNIE (文心) is unquestionably Baidu's model family, and Baidu documents the models at length, but no Baidu page asserts that a crawler by this name exists, gives it a string, or covers it in robots guidance; the consumer chat brand 文心一言 sits on the same model family, which is why the separately-listed `YiyanBot` may not be a separate fleet at all. Third-party directories describe a training fetcher for ERNIE: plausible, possibly accurate, not sourced to Baidu, and therefore not stated here as fact.
- Operated by
- Baidu
- Purpose
- Training
- robots.txt token
- ERNIEBot
- Verification
- User agent only
Collects pages into a corpus used to train models.
How to verify ERNIEBot
Baidu publishes no list of IP addresses for ERNIEBot, so nobody (us included) can prove that a request carrying this user agent really came from Baidu. 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 Baidu's own documentation on .
User agent
Baidu has not published a full user-agent string for ERNIEBot. Requests are identified by the ERNIEBot 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 ERNIEBot
Block
User-agent: ERNIEBot
Disallow: /Allow
User-agent: ERNIEBot
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 ERNIEBot costs you
Unknowable from primary sources, so the structural position is more useful than a guess. Baidu states that blocking the `Baiduspider` token blocks all of its user agents, which means a site that has already disallowed the parent has no separate lever left to pull here. A site that wants Baidu search but not Baidu model training has no vendor-documented way to express that, because the token that would carry the distinction has never been published. Adding a `User-agent: ERNIEBot` stanza costs nothing and may accomplish nothing: no commitment exists to honour it.
Vendor documentation
Baidu does not publish documentation for ERNIEBot that we could find. Everything on this page comes from what they do publish elsewhere and from observed behaviour, so treat it accordingly.