DeepL is one of the most popular machine translation engines among localization teams, and its API makes it easy to plug into almost any tool. Calling the API is simple. Making it part of a reliable translation workflow, with glossaries, tags, review and cost control, takes a bit more planning. This guide covers both.
What Is the DeepL API?
The DeepL API is a programmatic interface to DeepL's translation engine. Instead of pasting text into a web page, your software sends text or whole documents to DeepL and receives the translation back in seconds. It is used in content management systems, support tools, apps and, most often, inside translation management systems where it acts as one of several AI translation engines.
Free vs Pro API
DeepL offers a free API plan with a monthly character limit and paid Pro plans billed by usage. Every request is authenticated with an Authorization: DeepL-Auth-Key header. Check the official DeepL API documentation for current limits and pricing, as they change from time to time.
For production work, the Pro plan is usually the right choice. It removes the free character cap and gives you more control over data handling, which matters when you translate customer or confidential content.
Key Features That Matter for Localization
Glossaries
DeepL glossaries force specific translations for specific terms, such as product names or legal wording. This is the single most important feature for brand consistency. Keep the master list in your term base and sync it to the engine, rather than maintaining two separate lists.
Formality
For languages such as German, French or Spanish, you can ask for a more formal or more informal tone. Set this per project so that your app does not switch between "du" and "Sie" from one screen to the next.
Tag handling
When you send HTML or XML, DeepL can keep tags in place while translating the text around them. Placeholders such as {count} still need protection, so always run automated checks afterwards.
Document translation
The API can translate complete documents while keeping layout. This is useful for one-off files, but for ongoing localization it is better to work at segment level inside a CAT tool, where translation memory and review can be applied.
How to Use the DeepL API Inside a Translation Workflow
- Apply translation memory first. Only send segments to DeepL that have no exact or high fuzzy match in your translation memory. This can cut API usage substantially on repeat content.
- Attach the glossary. Send the relevant glossary with each request so key terms are always correct.
- Protect tags and placeholders. Use tag handling for markup and verify variables after translation.
- Send to post-editing by risk. Low-risk content can go straight to QA; customer-facing content should be reviewed by a linguist.
- Save approved translations. Every reviewed segment goes back to translation memory, so you never pay to translate it again.
Handling Errors and Limits
Production integrations need to handle failures gracefully. DeepL returns standard HTTP status codes. Rate-limit (429) and server (5xx) errors should be retried with exponential backoff. A 456 response means the character quota is used up, so retrying will not help; alert someone instead. Log usage per project so you can see which content types consume the most characters.
Is DeepL the Right Engine for Every Language?
DeepL performs very well for many European language pairs, but no engine wins everywhere. For some languages or content types, an LLM such as GPT or Claude may produce better tone or handle longer context more naturally. The best approach is to test engines on your own content. Our comparison of DeepL vs Google Translate and the guide on how to choose a machine translation tool explain what to measure.
Using DeepL in WordBeam
WordBeam connects DeepL alongside GPT, Claude and Grok, so you can choose the best engine per project and language without writing integration code. Translation memory is applied before any engine is called, your term base guides the output, and automated QA checks tags and placeholders. It is a complete machine translation software setup with human review built in.
Frequently Asked Questions
Is the DeepL API free?
There is a free API plan with a monthly character limit. For larger volumes and production use, DeepL offers paid Pro plans billed by usage.
Can I use DeepL glossaries through the API?
Yes. You can create glossaries for supported language pairs and reference them in translation requests.
Do I need to code to use the DeepL API?
Not if you use a TMS with a built-in DeepL connection. You add your settings once and translate from the editor.