AWS Translate cost: per character of text translated

Quick answer

Translate bills per character of input text (about $15 per million characters for standard translation), with a monthly free tier and higher rates for Active Custom Translation. Cost scales with text volume, so translating only what is needed, caching translations of repeated text, and avoiding re-translation are the levers.

Amazon Translate performs machine translation, billed by the number of characters in the input text. The model is simple, so cost tracks how much text you translate, and the common waste is re-translating the same content or translating text no one reads.

Per input character

Feature	Rate (approximate)
Standard translation	~$15 / million characters
Active Custom Translation	Higher per-character rate
Free tier	Monthly character allowance

Standard translation is about $15 per million input characters, with a free tier covering initial usage. Active Custom Translation (which adapts output using your parallel data) costs more per character. Custom terminology is available to enforce specific translations. Cost is characters times rate.

Cache and avoid re-translation

The biggest saving is not translating the same text twice. Caching translations, storing the translated output keyed by the source text and language pair, means repeated content (UI strings, common phrases, popular articles) is translated once and reused. For content that changes rarely, this eliminates most of the character volume.

Controlling Translate cost

Cache translations of repeated content, translate only text that will actually be read (do not pre-translate everything into every language speculatively), use standard translation unless custom adaptation is genuinely needed, and translate incrementally as content changes rather than re-translating in full. As with any per-unit service, sending less and reusing results are the levers, mirroring Comprehend.

FAQ

How is AWS Translate priced?

Per character of input text, about $15 per million characters for standard translation, with a monthly free tier and a higher rate for Active Custom Translation that adapts output using your parallel data. Cost scales with the volume of text you translate.

How do I reduce Translate cost?

Cache translations of repeated content so the same text is translated once and reused, translate only text that will actually be read rather than pre-translating everything into every language, use standard translation unless custom adaptation is needed, and translate incrementally as content changes.

Does caching help with Translate cost?

Significantly. Storing translated output keyed by source text and language pair means repeated content, UI strings, common phrases, popular articles, is translated once and reused, eliminating most character volume for content that changes rarely. Caching is the biggest Translate cost lever.

What is Active Custom Translation and does it cost more?

Active Custom Translation adapts translation output using your own parallel example data, improving domain-specific quality. It costs more per character than standard translation, so use it where the quality improvement justifies the higher rate, and standard translation elsewhere.

Should I pre-translate content into every language?

Only where it will be read. Speculatively translating all content into every supported language multiplies character volume for translations that may never be viewed. Translate on demand or for languages with actual audiences, and cache the results, rather than pre-translating everything.

Does C3X estimate Translate cost?

Translate cost is driven by characters translated, a usage input. C3X prices the surrounding infrastructure, and you model your translation volume to estimate the per-character charges.

What to do next

Estimate the infrastructure around your translation pipeline before you build it. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.