Open Paraphraser
Same meaning, different words, your choice of register.
AI Translator
Translate between thirteen languages without telling it what you pasted. The source language is worked out from the text, and you choose how close to the original the result should stay.
- Not stored by NexKit
- 20,000 characters
- No sign-up
Text to translate
or paste your own — anything goes
Settings
The source language is detected — you only choose the destination.
Approach
How far the translation may travel from the original wording.
⌘↵ to run
Translation
STANDING BY
This panel has nothing to do yet.
Put some text into 01 and it lands here. The sample works too, if you would rather not think of anything.
How your text is processed. Pressing Translate sends your text to NexKit's servers and on to a third-party AI provider to generate the output. NexKit does not store prompts or output after the request, and never links them to an account, because there are no accounts — that is not a guarantee about what happens on the provider's side, so avoid pasting confidential, sensitive or personal information. AI-generated output can be wrong; check anything that matters before you rely on it.
01Using AI Translator
- 01
Paste anything
You do not have to say what language it is in. That is worked out from the text itself.
- 02
Choose the destination
Then decide how close to stay: Faithful for accuracy, Natural for fluency, Localised to adapt units, dates and idioms too.
- 03
Translate
You get the translation and nothing else — no commentary, no notes about what it did.
What people use it for
- Read an email that arrived in a language you do not work in
- Send a reply that sounds like a native speaker wrote it
- Translate documentation without mangling the code samples
- Check whether a message needs the formal or informal register
Questions
How does it know what language I pasted?
The model identifies it from the text itself — there is no language to set and no detection step to wait for. If a translation ever comes back in the wrong language, the fix is to add a sentence of context; the tool is built to return the translation and nothing else.
What is the difference between Faithful, Natural and Localised?
Faithful stays near the source structure and prefers accuracy where the two conflict — the right choice for contracts and specifications. Natural reads as though written in the target language. Localised goes further and adapts units, currency, date order and idioms to the destination.
Will it translate names and code?
URLs, email addresses, code and numbers are left exactly as written every time — that is not a setting, so a translated README still contains runnable commands. "Keep proper nouns" decides the remaining case: with it on, names and brands stay as they are; with it off, a name that has a conventional form in the target language gets it.