Batch endpoints are good for retrieving multiple elements in one shot, also for large data retrieval and provides overall speed improvements.
The initial call is a bit slower, but all the subsequent requests when you paginate are fast and free. Its even cached for 24 hours.
As you pointed out if your results set is greater than 250 elements, paginating through the results is a batch endpoint feature.

Batch endpoints return all entries regardless if they are live/dead. I put a request on our development roadmap to add capabilities to filter for status, and other attributes.

Please explain when would I use batch endpoints? When I expect more results than 250? If I am looking only among the live/registered trademarks, is there a good way of only looking at the active tm. Thanks

Hello,

The search endpoints are optimzed for high throughput and are scaled for fast answer.

You can get the ALL trademark details including classification of a trademark by using the serial number endpoints.

A brief example of the classification attribute returned:

 {
  "international_code_total_no": "1",
  "international_code": "008",
  "us_code_total_no": "3",
  "us_code": "",
  "status_code": "6",
  "first_use_anywhere_date": "2021-09-15",
  "first_use_in_commerce_date": "2021-09-15",
  "primary_code": "008"
 }

This means you need to do in two steps, first use search to filter down dataset, then use serial endpoint for each to get details. You can use batch endpoints.

For more answers check out the FAQ.