Usage limits

Analyzed text limit

Natural Language Understanding truncates analyzed text that contains more than 50,000 single-byte or multibyte characters. To view the text that counts toward this limit in your requests, set the return_analyzed_text parameter to true.

You can set a smaller character limit with the limit_text_characters parameter. If you are interested in analyzing only a small portion of your content, this can help you avoid excessive costs.

Example parameters object:

{
  "url": "ibm.com",
  "limit_text_characters": 10000,
  "return_analyzed_text": true,
  "features": {
    "concepts": {}
  }
}

Request limit

Each Natural Language Understanding service instance is limited, based on the number of analyze requests being processed. For the Lite plan, the limit is 5 analyze requests per second; for the Standard plan, the limit is 150 analyze requests per second. Analyze requests exceeding these limits may receive a 429: Too Many Requests error.

The limit for the Standard plan can be increased by opening a support ticket.

All other requests are rate-limited at 5 requests per second.

Custom model training limit

The maximum number of models that can be trained via the Natural Language Understanding customization API in parallel is 3.

Custom model size limit for Lite pricing plans

There is a size limit for Knowledge Studio custom models that are deployed to Natural Language Understanding service instances on Lite pricing plans. To remove the custom model size limit, upgrade your Natural Language Understanding service instance to a paid pricing plan. You can find your service instances on the IBM Cloud resources page.

Language support

Different language restrictions apply depending on how you use the service. For details, see the Language support page.