demo · v138

Batch translator

Eight UI strings translated in one pass using a single Translator session. Reusing the session avoids the model-loading overhead on every call. Watch the strings resolve one by one and measure total time. Falls back to a simulated translation if the API is absent.

checking support…
One Translator session is created, then reused for all 8 strings. Session creation pays the model-load cost once; each subsequent translate() call costs only inference.
0 translated
8 total strings
— total ms
— ms / string
# Source (English) Translation Status

the code

const strings = ['Add to cart', 'Sign in', /* ... */];

// Create ONE session — pays model-load cost once
const translator = await Translator.create({
  sourceLanguage: 'en',
  targetLanguage: 'es',
});

// Reuse session for all strings — only inference cost each time
const results = await Promise.all(
  strings.map((s) => translator.translate(s))
);

translator.destroy(); // release model memory

see also