I have a project with 28k words (174k characters). I have connected DeepL Free API to the crowdin and I tried to translate the document via MT.
DeepL Free API limit is 500k characters.
I have pretranslated two languages without problem consuming about 348k characters from the limit.
Later I would like to translate a third language, which is obviously over the limit (348k + 174k = 522k, it is 22k characters over the free limit) so I hope that at least some text will be translated.
Crowdin started pretranslation but never came over 0% and failed without translation. Unfortunately, after this phase, DeepL limit was completely depleted.
The expected behavior would be to either not start the MT or translate at least something.