Hungarian. I’ve just tested it again, but I’m still not happy with the results.

Also, it always seems to suggest the word you’re translating in the “Alternatives” section for some reason. So, for instance, if I translate “moron” or “baffling”, DeepL thinks “moron” and “baffling” are valid alternatives. They’re not, these are neither Hungarian words nor English ones that a Hungarian would use.

But for quality, nuance, and privacy, DeepL is unmatched.

As much as I’d like this to be true, I have to disagree. Perhaps the quality of the translation depends on the language you translate from / to, but the results I got for my native language ranged from not very good to absolute rubbish.

What’s even more frustrating is that there’s no (easy) way for me to report such bad results so the developers could fix them. :/