How is this better than a basic translator?

An LLM translates for meaning and tone rather than word-for-word, so idioms and nuance survive. You can set the register, supply domain context for correct terminology, and get localisation notes explaining where a phrase had to be adapted because it has no direct equivalent.

What does the register control do?

Many languages distinguish formal and informal address — for example tu versus usted in Spanish or du versus Sie in German. Choosing formal or informal forces the right pronouns and politeness level; "match source" keeps whatever tone your original used.

Does it keep names and code intact?

Yes. The prompt instructs the model to leave names, numbers, code, and placeholders unchanged so you can translate UI strings or templated text without breaking variables. Always double-check placeholders survived on critical content.

Where does my API key go?

It stays in your browser tab and is sent directly to OpenAI or Anthropic with the request you trigger. It is never stored, logged, or routed through any Gera server, and refreshing the tab clears it.

Who pays for the API calls?

You do, on your own provider account. Each translation is one real API call billed at your usage rate. The tool itself is free — your only cost is the tokens consumed on your key.

AI Translation Tool (BYO Key)

Word-for-word translation loses the things that matter most — idiom, tone, and the right level of formality. This tool translates text into any of 18 languages using your own OpenAI or Anthropic key, preserving meaning and voice and explaining the choices it had to make, entirely in your browser.

How it works

Pick a provider and model, paste your API key, and choose the target language and register. Optionally name the domain — legal, marketing, medical, technical — so the model reaches for the correct terminology, and toggle whether you want localisation notes. Paste your text and the tool builds a prompt that asks for a natural, idiomatic translation that preserves tone, keeps names, numbers, code, and placeholders unchanged, and uses the register you specified. It makes one direct request to the provider and returns the translation, followed by localisation notes if requested.

For Anthropic, the request includes the official direct-browser-access header so it works straight from the page.

Getting accurate, natural results

Context is what separates a good translation from a literal one. If you are translating a marketing email, say so — the model will pick warmer, more persuasive phrasing than it would for a contract. The register control matters in languages with formal and informal address: getting tu versus usted or du versus Sie wrong can read as rude or oddly stiff. The localisation notes are worth keeping on for anything customer-facing, because they tell you which phrases were adapted rather than translated directly, so you can sanity-check the cultural calls.

Notes and limits

LLM translation is strong but not infallible, especially for low-resource languages, highly technical jargon, or text where a single mistranslation carries real consequences. For legal, medical, financial, or safety-critical material, treat the output as a fast first draft and have a fluent human reviewer confirm it before use. Always verify that placeholders and code survived intact when translating templated strings, and spot-check numbers and names on important documents.