Tokens for a production system

Hi,

First we would like to thank the LibreTranslate community fort the efforts provided to bring such a nice tool to the public!

We are a research group from University in UK working on Twitter data analytics for academic research. We manage a database of several hundred million tweets. We need to run your the LibreTranslate API live on our database to translate millions of non-English tweets everyday, in order to run sentiment analysis on them. We want to know what you think of this and to what extent you can help us?

Thank you,

1 Like

I’d definitely recommend to run your instance of LT on a server with lots of resources. Probably even multiple servers and a load balancer.

1 Like

Does LibreTranslate process separate requests on separate threads?

Yep. Due to Flask, each request is processed on a new thread.

1 Like

Thanks @pierotofy . I asked this question because of the github section where it’s stated that the code should be run only for test or infrequent use. Can I know what are the requests limitations of your API if there are any?

That section refers to the instance running on libretranslate.com. We wouldn’t be able to process the volume of requests you’re talking about. So you should run your own instance where you have no limitations (aside from those of the hardware).