How long can a single Extract API request take?

If a request reaches 180 seconds, it will be automatically timed-out and an error returned.

In practice: few if any requests will take anywhere close to this length of time, because Diffbot APIs have a default internal timeout  of only 30 seconds for the retrieval of third-party content. That is, the download and rendering of the requested web page can take a maximum of 30 seconds before being truncated and an error returned.

The subsequent Diffbot processing of a rendered page is measured in milliseconds, so only in rare exceptions can a response take more than 30 seconds.

For multipage articles that are automatically or manually concatenated, the 30-second fetch timeout applies to each individual page. So, for example, a site whose pages take an average of 15 seconds to load will incur, for a ten-page article, a total Diffbot API request time of ~150 seconds for a ten-page article.

The 30-second fetch timeout can be adjusted using the timeout argument in your Diffbot API request. E.g., a request argument of timeout=10000 will require a page to be fully retrieved in less than ten seconds; a request argument of timeout=60000 will extend this to one minute.