DeepL Pro Service Specification

1 DeepL Translator Pro

DeepL Translator Pro is an enhanced version of the DeepL translation service that provides Customer with a user interface through various front ends (e.g., website and desktop application) in accordance with the Service Specification outline herein. It allows Customer to utilize an improved interface to translate or refine texts, accessible via their DeepL Pro credentials on both Microsoft Windows and macOS applications or through a web browser. Customer on certain subscription plans may purchase Add-Ons for DeepL Translator Pro subscription. Depending on the specific Add-On, licenses may be allocated on an individual Internal User basis. If an Add-On includes a specific usage limit, the limit must be used within a one-month period and cannot be carried over to subsequent months.

The service enables the input of content for machine translations and supports the following set of source and target languages:

• Source language - Language of the text to be translated. Currently, the following languages are supported:

• Target language - The language into which you want to translate can be one of the following:

Within DeepL Translator Pro, Customer's texts and their translations will be processed as long as necessary for the performance of the Service.

For Customers opting to be charged on a usage basis for the Subscribed Services, Charges for texts entered into DeepL Translate Pro shall apply if the texts are translated or improved.

In addition, browsers, their JavaScript execution environment, and the HTTP transmission techniques used impose technical limits. DeepL Translator Pro will try to notify the Internal User about exceeding these limits and warn about possible performance issues.

DeepL Translator Pro allows Customer to translate whole documents. Ability to translate documents is subject to the ability of DeepL Translator Pro to correctly read and translate the documents, which can't be guaranteed.

Requirements and limitations for document translation:

• Number of Characters - Up to 1,000,000
• Number of files - The number of the files varies between the use of the website and the desktop applications.
  • Website - You can have one file translated for each translation process. You will not be able to select several files at the same time.
  • Desktop applications - You can have several files translated for each translation process. The files can be written in different source languages (as long as the said languages are supported by DeepL Translator). However, you are able to select only one target language for translation.
• Transferability of quotas - Monthly quotas may not be carried over to subsequent months.
• Transferability to other users - In the case of Single-User Licenses, quotas cannot be transferred to other Internal Users.

Included DeepL Voice allowance*

If and to the extent DeepL Translator Pro subscriptions include access to DeepL Voice, Customer’s use of DeepL Voice is subject to the Fair Usage Policy and section 3 of this Service Specification. Unused Audio Minutes expire at the end of each month and are not carried over to subsequent months. Once the monthly allowance is consumed, access to DeepL Voice is paused until the beginning of the following month. DeepL Translator Pro and any other subscribed Services remain unaffected. Audio Minutes cannot be transferred to other Internal Users. 

*DeepL reserves the right to change or remove the free of charge DeepL Voice allowance at its sole discretion.

DeepL MCP (Remote Model Context Protocol Server)

DeepL MCP is a hosted endpoint that enables Third-Party AI applications to access DeepL translation and writing functions via the Model Context Protocol ("MCP"). Compatible AI applications can use DeepL MCP to translate text and documents, improve text with DeepL Write (if applicable), and retrieve the languages, glossaries, style rules, and writing settings available to the user.

Access to DeepL MCP is available at mcp.deepl.com for Customers. 

The Internal User authorises the AI application by signing in with their DeepL account (OAuth 2.1). Access reflects the signed-in Internal User's individual account permissions.

DeepL MCP provides the following tools, equivalent to the corresponding DeepL functions:

Glossaries, style rules, and custom instructions cannot be created, modified, or deleted via DeepL MCP. Document translation uses a separate upload and download flow: document files are transferred directly between the Internal User's environment and DeepL and do not pass through the AI application.

Usage through DeepL MCP is attributed to the Customer's DeepL Pro Translator plan and counts towards the included usage volumes and applicable Fair Usage limits of the Customer’s plan.

Content submitted through DeepL MCP is processed by DeepL in the same way as direct use of the corresponding DeepL services. Before reaching DeepL, Content is processed by the AI application selected by the Customer. Such processing is governed by the Customer's own agreement with the provider of that AI application.

DeepL MCP is compatible with AI applications that support the MCP streamable HTTP transport, including Claude, ChatGPT, Microsoft Copilot, Cursor, and Continue. The availability, functionality, and conduct of compatible AI applications are outside DeepL's control and subject to change. The provisions regarding Third-Party Application usage in the agreement shall apply accordingly to any Third-Party Application used to access the Services via DeepL MCP.

DeepL MCP is currently not available for subscriptions with Data Residency Add-On.

2 DeepL Write Pro

DeepL Write Pro is an enhanced version of the DeepL text improvement and refinement service that provides Customer with a user interface through various front ends (e.g., website and desktop application) in accordance with the Service Specification outline herein. It allows Customer to utilize an improved interface to improve and refine texts, accessible via their DeepL Pro credentials on both Microsoft Windows and macOS applications or through a web browser.

DeepL Write Pro supports the following languages:

Within DeepL Write Pro, Customer's texts and their improved versions will be processed as long as necessary for the performance of the Service.

For Customers opting to be charged on a usage basis for the Subscribed Services, Charges for texts entered into DeepL Write Pro shall apply if the texts are translated or improved.

In addition, browsers, their JavaScript execution environment, and the HTTP transmission techniques used impose technical limits. DeepL Write Pro will try to notify the Internal User about exceeding these limits and warn about possible performance issues.

For DeepL Write Pro purchased as an Add-On, Customers may only purchase a number of Add-Ons equal to the number of User Subscriptions included in their DeepL Translator Pro subscription.

3 DeepL Voice

DeepL Voice includes DeepL for Conversations and DeepL for Meetings. The license to use DeepL Voice is a Single-User License. DeepL Voice can process requests more slowly or temporarily suspend the access to the Services on a case-by-case basis if the length of the meetings significantly exceeds the normal and reasonable use, and therefore the availability of the Services to all customers of DeepL may be affected.

In addition, browsers, their JavaScript execution environment, and the various transmission techniques used impose technical limits. DeepL Voice will try to notify the Internal User about exceeding these limits and warn about possible performance issues.

In addition to the terms and conditions of the Agreement, the following provisions shall also apply to DeepL Voice:

• Customer must obtain all necessary consents and authorisations from any third party, including Customer’s employees, for the transcription, translation and voice synthesis of the audio data by DeepL, which may be considered a recording under the respective laws applicable to Customer’s use of the service. In any way, Customer must ensure that any third parties invited to the meeting are aware that their audio data will be transcribed, translated and output by DeepL.
• Customer is aware that when using the text-to-speech and/or speech-to-speech functionality of the mobile device the native speech synthesis technology of the respective mobile device is used for this purpose. If the language is not supported, as a fallback EN-US (American English) is used. DeepL has no influence on the infrastructure and processing by native text-to-speech/speech-to-speech technology.
• Customer is not allowed to access and control DeepL Voice through automated procedures which may cause an increased number (exceeding usual human behaviour) of requests to DeepL Voice.

DeepL for Conversations

DeepL for Conversations is a bilingual conversation service that provides customers with real-time translations of face-to-face interactions. It features a user interface accessible through various front ends, in accordance with this Service Specification outlined herein. Customers can access DeepL Voice for Conversations using their designated access credentials via the iOS or Android DeepL applications. The service enables users to receive live transcription and translation of their multilingual conversations.

DeepL Voice for Conversations supports the following spoken languages (source by voice languages):

DeepL Voice for Conversations supports the following text languages (source by text languages):

DeepL Voice for Conversations supports one language pair per conversation.

DeepL Voice for Conversations supports the following translation languages (target languages):

Within DeepL Voice for Conversations, Customer's audio data, transcription data, and translation data will be processed as long as necessary for the performance of the Service.

DeepL Voice for Meetings

DeepL Voice for Meetings refers to a multilingual service that provides Customer with real-time transcription and translation of participant’s speech within virtual meetings.  As set out in this Service Specification DeepL Voice for Meetings also allows speech in a meeting to be output as synthesised speech. It offers a service accessible by Customer using their DeepL Voice for Meetings access credentials.

Via one or more access routes including but not limited to a desktop client or application the customer can access the service via a supported Third-Party Application on any platform supported by the respective Third-Party Application. The service allows Customer to receive a live translation and transcription of the meeting in the application and in a language of their choice.

DeepL Voice for Meetings is supported in the following Third-Party Applications:

• Microsoft Teams
• Zoom
• Google Meet

Users with an active DeepL Voice for Meetings license can initiate the generation of transcriptions and translations in a meeting. The manner in which this is accomplished on a technical level may vary depending on the Third-Party Application.

Users without an active DeepL Voice for Meetings license are still able to see content generated during a call if another license holder has initiated the generation of transcriptions and translations, however they will not be able to initiate the process.

Access to and visibility of the accompanying application for DeepL Voice for Meetings within the respective Third-Party Application’s ecosystem is controlled by the respective system’s access management system.

The DeepL desktop client / application for DeepL Voice for Meetings is needed to use the speech-to-speech functionality and to start/host meetings. Where DeepL provides a desktop client / application, DeepL grants Customer a non-exclusive, non-sublicensable right, limited to the term of the Agreement, to install and use the desktop client/application on devices of the holder of the Single User License. Only users with an active license can benefit from synthesised speech output generated during a meeting. Users without an active license are still able to see content generated during a meeting if another license holder has initiated the generation of transcriptions and translations, however they will not be able to initiate the process.

DeepL Voice for Meetings supports the following spoken languages (source languages):

DeepL Voice for Meetings supports multiple spoken languages per meeting, such that every meeting participant can speak their own language.

DeepL Voice for Meetings supports the following translation languages (target languages):

Within DeepL Voice for Meetings, Customer's audio data, transcription data, and translation data will be processed as long as necessary for the performance of the Service.

In addition to the terms and conditions of the Agreement, the following provisions shall also apply to DeepL Voice for Meetings:

In the event that (a) a Third-Party Application deprecates support of DeepL Voice for Meetings; or (b) a change to a Third-Party Application requires changes to DeepL Voice for Meetings which, considering the interests of both Parties, cannot be implemented in a technically or economically reasonable manner by DeepL, both Parties may terminate this Agreement with regard to DeepL Voice for Meetings without notice.

4 DeepL API

The DeepL API is accessible through a REST interface. Request parameters are used to pass information to the DeepL API, results are returned in a suitable representation (e.g., JSON). Request parameters must be encoded in accordance with the documentation provided under https://developers.deepl.com/docs. Authentication is based on fixed keys, SSL handling transport security. The translate function provides the following functionality:

• Text - Text to be translated.
  • Only text encoded in accordance with the documentation is supported.
• Source language - Language of the text to be translated. Currently, the following languages are supported:

If this parameter is omitted, the DeepL API will make efforts to detect the language of the text.

• Target language - The language into which you want to translate can be one of the following:

The translate function returns the following representation of the processing result:

• Detected source language - the language which has been detected for the source text.
• Text - the translated text.

The DeepL API is designed to translate at least 600 characters per minute and per Customer; exceeding use may be limited, and Customers may encounter a 429 “Too Many Requests” error message, as described in the documentation.

Applications using the DeepL API should implement a mechanism to handle such responses accordingly and, if appropriate, to try again later. A mechanism increasing the delay for another request exponentially is recommended.

The DeepL API allows Customer to translate whole documents. Ability to translate documents is subject to the ability of the DeepL API to correctly read and translate the documents, which can't be guaranteed.

Requirements and limitations for document translation:

• File formats and file sizes

• Number of Characters - Up to 1,000,000

For document translations, a minimum of 50,000 Characters per document is charged.

DeepL MCP for API (Remote Model Context Protocol Server)

If and to the extent DeepL MCP use is included in an API plan, usage of the Services via DeepL MCP authenticated with Customer’s API key constitutes use of the API. In that case the DeepL API can be accessed via DeepL's hosted Model Context Protocol endpoint at mcp-api.deepl.com, using the same API key as api.deepl.com.

DeepL MCP provides the following tools, equivalent to the corresponding DeepL functions:

Glossaries, style rules, and custom instructions cannot be created, modified, or deleted via DeepL MCP. Document translation uses a separate upload and download flow: document files are transferred directly between the Internal User's environment and DeepL and do not pass through the AI application.

Usage through DeepL MCP is attributed to the Customer's DeepL API Pro subscription and counts towards the included usage volumes and applicable Fair Usage limits of the Customer’s subscription plan. 

Content submitted through DeepL MCP is processed by DeepL in the same way as direct use of the corresponding DeepL services. Before reaching DeepL, Content is processed by the AI application selected by the Customer. Such processing is governed by the Customer's own agreement with the provider of that AI application.

DeepL MCP is compatible with AI applications that support the MCP streamable HTTP transport, including Claude, ChatGPT, Microsoft Copilot, Cursor, and Continue. The availability, functionality, and conduct of compatible AI applications are outside DeepL's control and subject to change. DeepL is not responsible for Third-Party AI Applications and shall have no liability in relation to the processing of Content by such applications.

DeepL MCP is currently not available for subscriptions with Data Residency Add-On.

DeepL Write API

The DeepL Pro API (v2 endpoint only) also includes a text improvement function. This function is only available to users with a DeepL API Pro subscription type.

The text improvement function provides the following functionality:

• Text - Text to be improved.
  • Only text encoded in accordance with the documentation is supported.
• Target language - The language in which your improvements are provided can be one of the following:

• Text - The improved text.
• Target language - The target language specified by the user.
• Detected source language - The detected source language of the text provided in the request.

DeepL Voice API

The DeepL API Pro (v3 endpoint only) also includes real-time audio translation. It provides two functions over a real-time audio stream: speech-to-text (translated text output) and speech-to-speech (translated speech output). Both functions are only available to users whose DeepL API subscription type includes them.  Speech-to-speech is additionally limited to eligible plans. DeepL retains the right to introduce a concurrent session limit.

Speech-to-text

• Audio Stream - Realtime Audio stream to be translated into text in up to 5 languages.
• Source Languages (Audio) - The realtime audio stream you want to have translated can be a single source language of the following list:

• Target language (Text) - The language in which your translations are provided can be one of the following:

The speech to text function returns the following representation of the processing result:

• Language - The language explicitly set or automatically detected for your audio.
• Text - The translated text(s) and the transcribed source as a text.

Speech-to-speech

The speech-to-speech function for real-time audio provides the following functionality:

• Audio Stream - a real-time audio stream translated into synthesized speech in one language. Source and target text transcripts are returned alongside the translated speech, subject to the speech-to-text target language limits below.
• Source Languages (Audio) - The realtime audio stream you want to have translated can be in one of the following languages:

• Target language (Speech) - The language in which your translations are provided can be one of the following:

The speech-to-speech function returns the following representation of the processing result:

• Language - The language detected for your audio and the requested speech output language.
• Text - The translated text and the transcribed source as a text.
• Speech - the synthesized translated audio for the target language.

The DeepL API Pro for speech-to-text and speech-to-speech is designed to process a specific amount of target languages per stream for a specific maximum streaming connection duration and audio chunk size. The exact applicable information and units can be found in the Documentation. The audio stream speed may be higher than real-time, but not higher than the value provided in the Documentation; exceeding use may be limited, and Customers may encounter a 429 “Too Many Requests” error message, as described in the documentation.

Applications using the DeepL API Pro should implement a mechanism to handle such responses accordingly and, if appropriate, to try again later. A mechanism increasing the delay for another request exponentially is recommended.

5 Translation Flows

Translation Flows is an end-to-end translation workflow service that enables Customer to import content from supported third-party platforms, translate it using DeepL's machine translation technology, manage translation jobs, and deliver translated content back to the source platform. The service is accessible via the DeepL web application using Customer's DeepL Pro credentials.

Supported Integrations

DeepL AI-Flows supports content import from and delivery to the following third-party platforms:

• Contentful
• Google Drive

The availability of integrations is subject to change. DeepL may add support for additional third-party platforms over time.

Supported File Formats

When importing documents via a supported integration, the following file formats are supported for translation: