OneTag AI

Free translation API for websites: no backend, no credit card

Is there a free translation API I can call straight from a web page?

Yes. OneTag AI translates text with a language model, and each email address gets a key with 1,000,000 free tokens a month and no credit card. In our test that covered about 1.9 million characters of English a month, roughly 300,000 words. The key is safe to leave in page source, so a static site can translate text straight from the browser with no backend.

The allowance starts again on the 1st of each month (UTC). This page has code for one string, for many strings at once and for curl, the numbers we measured, and a comparison with the other free translation APIs. Two of them are a better pick in some cases, and the page says which.

Steps

1. Get a free key

   Enter your email at onetagai.com. The key (it starts with ot_) arrives by email. One key per email, no card.

2. Translate one string

   Add the script tag, name the target language in plain words, and pass the text:

   <script src="https://onetagai.com/v1.js" data-key="YOUR_ONETAG_KEY"></script>
   <script type="module">
     const spanish = await onetag.ask(
       "Translate to Spanish. Return only the translation.",
       "Your order has shipped and will arrive on Friday."
     );
     console.log(spanish);
   </script>

   In our test this printed Tu pedido ha sido enviado y llegará el viernes. after about 2.5 seconds and used 90 tokens. The source language is detected by the model; you only name the target.

3. Translate many strings in one call

   For interface text, send the strings as one JSON array and ask for them back under a key, so the pieces stay aligned:

   <script type="module">
     const strings = ["Add to cart", "Free shipping on orders over $50", "Sign in"];
     const { t } = await onetag.ask(
       'Translate every string in this JSON array to Japanese. Return JSON {"t": [the translations, same length, same order]}.',
       JSON.stringify(strings),
       { json: true, maxTokens: 2000 }
     );
     console.log(t);
   </script>

   t is an array of the same length and order. With json: true the answer is always a JSON object, which is why the prompt asks for the array under t.

4. Or call it over HTTP

   The same key works from curl or any server language. Set max_tokens, because the default answer length is 500 tokens:

   curl https://onetagai.com/v1/ask \
     -H "Content-Type: application/json" \
     -d '{"key": "YOUR_ONETAG_KEY", "prompt": "Translate to German. Return only the translation.", "input": "Where is the nearest train station?", "max_tokens": 1000}'

   The reply is {"text": "Wo ist der nächste Bahnhof?", "usage": {"input": 76, "output": 6}}.

How much the free key translates

Tokens count the text you send plus the translation that comes back. We translated a 1,929-character English text on 2026-11-10:

So 1,000,000 tokens is about 1.9 million characters of English source text a month. A single sentence comes back in 2 to 3 seconds. A long text takes several seconds, so pass a stream callback to show the translation while it is written.

Free translation APIs compared

Translation APIs were built to be called from a server, with a secret key. On a website the first question is whether the page itself can call the API.

Other providers' figures were read on their own pages on 2026-11-10; follow the link in each row for the current numbers. Free tiers change often.

Which free translation API should you use?

• A few short strings, and you want no sign-up at all: MyMemory. No key, callable from the page, 5,000 characters a day and 500 bytes per request.
• A static site that translates paragraphs, pages or user text: OneTag AI. One public key, up to 32,000 characters per call, and the same tag also summarizes, answers questions and reads images.
• You have a server and want a dedicated translation engine with glossaries or document translation: DeepL, Google Cloud Translation or Azure Translator. Azure's free tier is the largest of the three.
• Text must not leave your own machines: LibreTranslate, self-hosted.

Limits to know before you rely on it

• It is a language model, not a dedicated translation engine. There are no glossaries and no language-code parameters: you name the language in the prompt.
• The same text can be worded slightly differently on two calls. Cache the result if the wording has to stay fixed.
• One call takes up to 32,000 characters of text. Answers are 500 tokens long by default and up to 10,000 with maxTokens (max_tokens over HTTP); a translation that stops halfway means the limit was too low.
• 60 requests a minute per key and 20 a minute per IP address.
• Good for interface text, help articles and messages. For contracts or medical text, have a person check it.

Translate a whole page

To switch every text node of a page to the visitor's language and back, with a language menu, see Website translator widget in JavaScript.

Questions

Related guides

• Website translator widget in JavaScript: a Google Translate widget alternative
• Free AI API for websites: no backend, no credit card
• Free OCR API: image to text from a web page, no backend
• Free chatbot API for websites: chat history, streaming, no backend

Updated 2026-11-11. Something wrong or missing? Email support@odysseyx.app.