# Translate a message or short text

> Translate a message or short text: 0 companies fully do it, 14 with a limit, 17 carry the tag. Every listing in Translation with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/translation.translate-text/
- Markdown: https://www.anchorterminal.com/niches/translation.translate-text/index.md (~1,650 tokens)
- Slim: https://www.anchorterminal.com/niches/translation.translate-text/index.min.md (~380 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/translation.translate-text/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 14 | 17 | 19 | 12 |

## What counts as done

The text in the target language with nothing added or dropped, the detected source language returned, and a clear error or flag rather than the source echoed back when translation fails.

Measured, for our own tests: one call under 2 seconds for 1,000 characters.

How agents ask for it: translate this email into Spanish; what does this comment say; reply to this customer in Japanese; translate the text in this screenshot

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Amazon Translate (Amazon Web Services, Inc.)**: TranslateText takes SourceLanguageCode auto (via Comprehend), returns the source code and raises DetectedLanguageLowConfidenceException, up to 10,000 bytes a call. A first-hand report saw a bare word echoed back as a success.
- **Azure Translator (Microsoft Corporation)**: Translate takes 1,000 elements and 50,000 characters a request, detects and returns the source, NMT or LLM per request; the pages read don't say what flags an untranslated result, and a first-hand plan falls back to an LLM where Azure lacks a language.
- **Crowdin (Crowdin OÜ)**: Translate via MT takes up to 100 strings a call through an engine configured in the project, with languageRecognitionProvider for detection and the source language in the response; no untranslated flag is documented.
- **DeepL API (DeepL SE)**: /v2/translate returns detected_source_language with each text and takes context, formality and glossaries, within 128 KiB a request. Open first-hand reports include a wrong detected source (deepl-node #57); no untranslated flag is documented.
- **Google Cloud Translation (Google LLC)**: translateText detects the source when none is given and returns detectedLanguageCode, with 5,000 code points recommended per request and 6,000 v3 requests a minute, but no untranslated flag is documented.
- **Lara Translate API (Translated s.r.l.)**: translate() takes a string, array or TextBlocks with instructions, memories and glossaries, and returns source_language when detection is used, but no untranslated flag or failure status per segment is documented, and the free plan is 10,000 characters a month.
- **LibreTranslate (LibreTranslate)**: /translate takes source auto and returns detectedLanguage with a confidence, but open first-hand reports show sentences returned in English with no error and short mixed English-Chinese strings gaining boilerplate and changed numbers.
- **Lingo.dev (Replexica, Inc.)**: Localize translates key-value data synchronously through an engine's glossary and brand voice, with per-key hints, and a separate recognize endpoint detects language. sourceLocale is required on Localize, and an open report shows strings written in Spanish instead of the target.
- **Lingvanex Translation API (Nordicwise Limited)**: /translate takes a string or array with auto-detection when from is omitted and an err field on failure, but the spec does not show the detected source in the translate response and only 200 and 403 are documented.
- **Localazy (Localazy s.r.o.)**: The AI Translation endpoint (beta) translates several items, plural forms included, with context, style guide and glossary, but from is required, so there is no detection, and no untranslated flag is documented.
- **Phrase (Phrase a.s.)**: /v2/textTranslations takes up to six texts with sourceLang optional and returns sourceLang, through engines chosen per pair, but texts are capped at 2,000 characters, no untranslated flag is documented and the rate limit is unpublished.
- **Smartcat (Smartcat Platform Inc.)**: The real-time API returns a translation for a piece of text, but only through a smart translation profile a person creates in the web app first, and detection and untranslated flags are not documented; API access is listed from the Anticipate plan.
- **SYSTRAN Translate API (SYSTRAN SAS)**: /translation/text/translate takes repeated input, source auto, and withInfo returns language identification with a confidence plus a per-output error field, but the terms bar automated access beyond normal human volumes and no rate limit is documented.
- **TranslatePlus (PetaBytes, Inc.)**: /v2/translate takes up to 5,000 characters with source auto and returns text, translation, source and target, but no untranslated flag is documented, and the Free plan caps each request at 500 characters.

- not this job: Lokalise (Lokalise, Inc.): No synchronous translate call; machine and AI translation run as automatic_translation tasks over keys in a project.
- not this job: POEditor MCP server (Code Whale Inc.): No free-text translate tool; automatic translation runs over a project's terms with Google, Microsoft or DeepL.

## What stops agents finishing

- source text returned unchanged with no error
- target language not offered, so the agent needs a second provider
- detection wrong on short text so the wrong pair is used

## Around it

Before: email.read-message, support.read-ticket, speech-to-text.transcribe, document-extraction.ocr. After: email.send-reply, support.post-reply, messaging.send.

## Where this comes from

Weight 19 of 100 in Translation: 12 of 55 accounts (silent untranslated output, added text, wrong detection, low-resource languages)

As data: https://www.anchorterminal.com/api/v1/use-cases/translation.translate-text.json, or the MCP tool get_use_case with id translation.translate-text. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
