Multilingual Translation

Translates text between 28 languages, one text or a batch; none of 13,156 test translations in 13 directions failed.

POST/v1/text/translate

+1 endpoints

Try it

• Price1 ICper 400 characters
• MeasuredchrF++ 56.1
• Served4.2Mcalls

• v1.0.4-20261028
• Updated 28 Oct 2026
• Active
• Benchmark paper (PDF)

Input

text

Hello

source_lang

en

target_lang

th

OutputPOST /v1/text/translate

translation

สวัสดี

processing_time

0.056

Every field, under ResponseOther endpoints: /batch_translate see Request

Request

curl -X POST "https://api.iapp.co.th/v1/text/translate" \
  -H "apikey: YOUR_API_KEY" \
  -F "text=Hello" \
  -F "source_lang=en" \
  -F "target_lang=th"

Response

200OK. The fields are explained under Response.

{
    "translation": "สวัสดี",
    "processing_time": 0.056
}

Request

Form fields for one text, JSON for a batch. 1 IC per 400 characters.

Headers

apikeystringrequired

Your API key, created under API Keys in the control panel.

Body

textstringrequired

The text to translate.

source_langstringrequired

The source, as a code: ar, bn, cs, de, en, es, fa, fr, he, hi, id, it, ja, km, ko, lo, ms, my, nl, pl, pt, ru, th, tl, tr, ur, vi or zh.

target_langstringrequired

The target, one of the same codes.

max_lengthinteger

The most output tokens.

Endpoint	Body
/v1/text/translate	The fields above
/v1/text/batch_translate	An array of {text, source_lang, target_lang}

/v3/store/nlp/multilingual-translation and …/batch_translate are the same endpoints under the store prefix.

Response

One text: translation and processing_time, as in the panel. A batch: results[], each with translation and the index of its input.

One text: translation and processing_time, as in the panel. A batch: results[], each with translation and the index of its input.

Details

Measured

On the production service, 28 October 2026, against the public FLORES-200 benchmark: the same 1,012 professionally translated sentences in each of 13 directions, 13,156 translations, none failed. chrF++ measures how close a translation is to a professional reference translation, from 0 to 100.

Direction	Result
Thai to English	chrF++ 56.1
English to Thai	chrF++ 46.6
Chinese / Japanese to English	chrF++ 55.5 / 54.0
Chinese / Korean / Japanese to Thai	chrF++ 40.1 / 39.8 / 38.7
Thai to Chinese / Japanese / Korean	chrF++ 23.7 / 26.3 / 28.4
English / Thai to Burmese	chrF++ 29.2 / 27.4; fragments of unrelated languages in 55.4% / 53.0% of answers
Answers left in the source language, all directions	48 of 13,156

Time per sentence, measured on 28 Oct 2026 through the public API with 40 Thai sentences of the FLORES-200 benchmark: 0.10 s in batches of 20, 0.70 s one at a time, network included.

Compare scores only within one target language: chrF++ runs lower for Chinese and Japanese, which do not separate words. Burmese output is not reliable and should be reviewed before use; in every other direction at most 1.1% of answers contain such fragments. Method, all 13 directions and confidence intervals: the benchmark (PDF).

Limits

• Only the languages listed under source_lang.
• Burmese output is not reliable: review it before use (see Measured).
• No limit on input length: long texts are split at sentence boundaries.
• For volume, use the batch endpoint: it translates up to eight texts at the same time, 86% less time per sentence than single requests.

Data handling

Text is processed in memory and not kept after the response. Consent and notice under PDPA and GDPR remain with you as the data controller.

Changelog

Version	Date	Change
v1.0.4-20261028	28 Oct 2026	/batch_translate translates up to eight texts at the same time: 0.10 s per sentence end to end in batches of 20, against 0.56 s before. Every request is billed at the listed rate of 1 IC per 400 characters, batches included.