FAQ

Frequently Asked Questions about 638Labs API/services.

Does 638Labs.com support streaming?

Yes (early access), 638Labs API supports streaming for all endpoints. Simply set the stream=true parameter in your payload request. Important, this is in the payload JSON ​

Does 638Labs support batching?

Yes, 638Labs supports batching for all models/agents. Please contact us for details - if there is a use case/provider that we have not integrated with yet, contact us and let us know. ​

Does 638Labs support custom models/agents/knowledgebases?

Yes. For the self-serve accounts the target API call must be OpenAI Compatible. For custom ends that require any special authentication handshake before making an API call, please contact us.

If you have a specific use case, please contact us. ​

Does 638Labs use my data?

No, 638Labs does not use your data for training, or any other purpose. Usage data is scheduled for deletion after 30 days.

What is the difference between datasources and knowledgebases?

For the purposes of this service, there is no difference. They both represent an endpoint that we can access via an HTTP POST request. The “datasource” or “knowledgebase” endpoint is expected to 1) store data, 2) fetch this data and 3) return it in a format your upstream systems can consume it in.