DeepL vs. Google Translate: data protection per plan

no: according to q1 section 3, no personal data may be translated in the free service; processing of personal data is only possible under a DeepL Pro subscription (q1). No DPA documented for Free.

no: Terms and conditions: processing of personal data with the Free API prohibited; in the event of a breach, the customer is obliged to indemnify (q2); privacy policy section 13 refers to the terms and conditions (q1).

yes: Terms and conditions 8.1.5 refer to the DeepL Data Processing Agreement in the Trust Center, which becomes 'an integral part of this Agreement'; privacy policy section 4: concluded on request to privacy@deepl.com (q1, q2)

yes: Terms and conditions 8.1.5 refer to the DeepL Data Processing Agreement in the Trust Center, which becomes 'an integral part of this Agreement'; privacy policy section 4: concluded on request to privacy@deepl.com (q1, q2)

yes: Terms and conditions 8.1.5 refer to the DeepL Data Processing Agreement in the Trust Center, which becomes 'an integral part of this Agreement'; privacy policy section 4: concluded on request to privacy@deepl.com (q1, q2)

yes: Terms and conditions 8.1.5 refer to the DeepL Data Processing Agreement in the Trust Center, which becomes 'an integral part of this Agreement'; privacy policy section 4: concluded on request to privacy@deepl.com (q1, q2)

yes: Terms and conditions 8.1.5 refer to the DeepL Data Processing Agreement in the Trust Center, which becomes 'an integral part of this Agreement'; privacy policy section 4: concluded on request to privacy@deepl.com (q1, q2)

not publicly documented (consumer service under the Google Terms of Service; in Google Workspace, too, Google Translate is not a Core Service but an additional service for which Workspace Help names the Google Terms of Service, q9; no DPA is mentioned)

yes (data usage FAQ: 'Cloud Translation API complies with the Cloud Data Processing Addendum', q2, q4)

yes (data usage FAQ: 'Cloud Translation API complies with the Cloud Data Processing Addendum'; according to the FAQ, glossaries are also stored in accordance with the CDPA, q2, q4)

"We process the content you upload and their translations or improvements for a limited period of time to train and improve our neural networks and algorithms. This also applies to corrections you make to our translation or improvement suggestions." (q1, section 3). In addition: personal data in the free service "may be checked by a human" (q1, only in the note on the Korean version, section 24.2).

q1 section 13 refers, for "DeepL API Free", to the terms and conditions and to "section 3 on using the free version of DeepL" (section 3 describes use for training, see the Free plan). It contains no express statement on training specifically for API Free. Developer documentation q8: "With DeepL API paid plans, texts aren't saved on persistent storage and aren't used to train our models"; this expressly applies only to paid plans. The pricing page (q10, API tab) lists the free API plan "Developer" (one-off credit of 1 million characters) and names "no use of your data for model training" (translated) only for the Growth plan ("all the benefits of Developer, plus:", translated); whether Developer is identical to API Free is not stated there.

"When using these services, your texts will not be used to improve the quality of our services." (q1, section 4, applies to DeepL Translator Pro, DeepL API Pro, DeepL Write Pro). Pricing page: "No data training" (q3). Terms and conditions 7.4: rights to Content remain with the customer; DeepL may use Content "solely in order to provide DeepL's Services" (q2).

"When using these services, your texts will not be used to improve the quality of our services." (q1, section 4, applies to DeepL Translator Pro, DeepL API Pro, DeepL Write Pro). Pricing page: "No data training" (q3). Terms and conditions 7.4: rights to Content remain with the customer; DeepL may use Content "solely in order to provide DeepL's Services" (q2).

"When using these services, your texts will not be used to improve the quality of our services." (q1, section 4, applies to DeepL Translator Pro, DeepL API Pro, DeepL Write Pro). Pricing page: "No data training" (q3). Terms and conditions 7.4: rights to Content remain with the customer; DeepL may use Content "solely in order to provide DeepL's Services" (q2).

"When using these services, your texts will not be used to improve the quality of our services." (q1, section 4, applies to DeepL Translator Pro, DeepL API Pro, DeepL Write Pro). Pricing page: "No data training" (q3). Terms and conditions 7.4: rights to Content remain with the customer; DeepL may use Content "solely in order to provide DeepL's Services" (q2). The pricing page additionally names Bring Your Own Key (q3, q4).

"When using these services, your texts will not be used to improve the quality of our services." (q1, section 4, applies to DeepL Translator Pro, DeepL API Pro, DeepL Write Pro). Pricing page: "No data training" (q3). Terms and conditions 7.4: rights to Content remain with the customer; DeepL may use Content "solely in order to provide DeepL's Services" (q2). Developer documentation: "With DeepL API paid plans, texts aren't saved on persistent storage and aren't used to train our models." (q8).

A Translate-specific statement on the use of entered texts for model training and an opt-out are not documented. The Google privacy policy (effective from 2026-10-16) states: 'We analyze usage of Google Translate to improve translation quality and increase the availability of Translate in more languages.' as well as the analysis of how often translations are marked as poor quality; publicly available information is used to train models for products such as Google Translate (q6).

Data usage FAQ (as of 2026-10-13): 'Google does not use any of your content for any purpose except to provide you with the Cloud Translation API service.' and 'Google does not use the content you send to train and improve our Google Translation features.' Content is not shared with third parties (q2).

Data usage FAQ (as of 2026-10-13): 'Google does not use any of your content for any purpose except to provide you with the Cloud Translation API service.' and 'Google does not use the content you send to train and improve our Google Translation features.' (q2). Custom models are trained with the customer's training data in the customer's project (pricing page: training billed by the hour, q1).

No selectable location option documented for Free. According to the Trust Center document 'Infrastructure Update' (q5), proprietary DeepL data centres in Iceland and Sweden are used for research, model training and serving Free-tier users. Privacy policy (q1, section 7.1): document translations (Free and Pro) are "processed on our infrastructure in the EEA".

Developer documentation q7: 'Data handling may differ between the Free and Pro plans.' The location for API Free is not stated; q5 refers to Free-tier users in general (proprietary data centres in Iceland/Sweden) without expressly mentioning API Free.

Terms and conditions 3.1.3: DeepL determines the place of processing and may use third-party cloud services; the place of processing 'may depend, among other things, on the location of the customer' (q2). The Trust Center document (q5) names AWS for 'business and enterprise customers'; whether the Individual plan falls under this is not stated there. The pricing page describes no location for Individual (q3). Trust Center FAQ (q5): "Self-serve customers signing up for a subscription from 13 February 2025 may be processed on DeepL administered servers as well as on AWS servers."; the FAQ does not name the AWS regions for self-service subscriptions.

As Pro – Individual: no location on the pricing page (q3); the Trust Center document names AWS for 'business and enterprise customers' without attributing the Team plan (q5); terms and conditions 3.1.3 (q2). Trust Center FAQ (q5): "Self-serve customers signing up for a subscription from 13 February 2025 may be processed on DeepL administered servers as well as on AWS servers."; the FAQ does not name the AWS regions for self-service subscriptions.

According to the Trust Center document 'Infrastructure Update' (q5): hybrid model of proprietary data centres (Iceland/Sweden; research, model training, Free tier) and AWS, which takes over 'data processing and storage for business and enterprise customers'. Default: customer content may be processed and stored in all AWS regions (currently EU, US, JP); non-content metadata is stored centrally. The paid 'Data Residency' add-on (for sales-assisted customers) contractually guarantees processing only in the selected region (EU, US or JP). Terms and conditions 3.1.3: DeepL determines the place of processing; third-party cloud services may be used (q2).

According to the Trust Center document 'Infrastructure Update' (q5): hybrid model of proprietary data centres (Iceland/Sweden; research, model training, Free tier) and AWS, which takes over 'data processing and storage for business and enterprise customers'. Default: customer content may be processed and stored in all AWS regions (currently EU, US, JP); non-content metadata is stored centrally. The paid 'Data Residency' add-on (for sales-assisted customers) contractually guarantees processing only in the selected region (EU, US or JP). Terms and conditions 3.1.3: DeepL determines the place of processing; third-party cloud services may be used (q2). For Enterprise customers, according to q5, the residency option can be booked contractually (EU, US, JP).

Two statements, verbatim: (a) Trust Center document (q5): the default is global processing via the AWS regions EU, US, JP; Data Residency add-on for sales-assisted customers. (b) Developer documentation (q6): regional endpoints process and store data in the region: 'European Union (default)' api.deepl.com, 'United States' api-us.deepl.com, 'Japan' api-jp.deepl.com; US/JP only with a signed 'regional deployment addendum' (otherwise HTTP 403). The relationship between (a) and (b) is not explained in the sources.

Google privacy policy: 'We maintain servers around the world and your information may be processed on servers located outside of the country where you live.' A Translate-specific location statement or region selection is not documented (q6).

Data usage FAQ: Cloud Translation – Basic has global endpoints (regional endpoints are documented only for Advanced, q3); 'the Services are global and data cannot be configured to remain in a specific region for processing' (q2). CDPA: subject to agreed location commitments, customer data may be processed in any country in which Google or its sub-processors maintain facilities (q4).

Documentation 'Global and multi-regional endpoints' (as of 2026-10-13): besides the global endpoint there are translate-eu.googleapis.com and translate-us.googleapis.com; with a multi-regional endpoint, data at rest and ML processing stay within the EU or the US respectively. The default is the global endpoint. Custom AutoML models and Preview features are not supported on the multi-regional endpoints (only the pre-trained NMT model). Organisation policy administrators can restrict the regions available for Advanced resources via a resource locations constraint (q3). The data usage FAQ refers to the 'AI/ML Data Location' section of the Service Specific Terms for regional endpoints (q2).

• not publicly documented: full list only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5). q5 names proprietary DeepL data centres (Iceland/Sweden) instead of AWS for Free-tier users.

• not publicly documented: full list only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5). q5 names proprietary DeepL data centres (Iceland/Sweden) instead of AWS for Free-tier users.

• Amazon Web Services (AWS): according to q5 ('Infrastructure Update'), data processing and storage for Business and Enterprise customers; regions EU, US, JP
• ElevenLabs: only for DeepL Voice (substitute voices), according to q1 section 5
• Microsoft Ireland or AWS EMEA: only for DeepL Voice for Meetings (hosting of the bot application, forwarding of audio data to DeepL infrastructure; transcription/translation not on their servers), according to q1 section 5
• Full sub-processor list: only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5/q9); the entries listed here are only those expressly named in q1/q5

• Amazon Web Services (AWS): according to q5 ('Infrastructure Update'), data processing and storage for Business and Enterprise customers; regions EU, US, JP
• ElevenLabs: only for DeepL Voice (substitute voices), according to q1 section 5
• Microsoft Ireland or AWS EMEA: only for DeepL Voice for Meetings (hosting of the bot application, forwarding of audio data to DeepL infrastructure; transcription/translation not on their servers), according to q1 section 5
• Full sub-processor list: only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5/q9); the entries listed here are only those expressly named in q1/q5

• Amazon Web Services (AWS): according to q5 ('Infrastructure Update'), data processing and storage for Business and Enterprise customers; regions EU, US, JP
• ElevenLabs: only for DeepL Voice (substitute voices), according to q1 section 5
• Microsoft Ireland or AWS EMEA: only for DeepL Voice for Meetings (hosting of the bot application, forwarding of audio data to DeepL infrastructure; transcription/translation not on their servers), according to q1 section 5
• Full sub-processor list: only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5/q9); the entries listed here are only those expressly named in q1/q5

• Amazon Web Services (AWS): according to q5 ('Infrastructure Update'), data processing and storage for Business and Enterprise customers; regions EU, US, JP
• ElevenLabs: only for DeepL Voice (substitute voices), according to q1 section 5
• Microsoft Ireland or AWS EMEA: only for DeepL Voice for Meetings (hosting of the bot application, forwarding of audio data to DeepL infrastructure; transcription/translation not on their servers), according to q1 section 5
• Full sub-processor list: only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5/q9); the entries listed here are only those expressly named in q1/q5

• Amazon Web Services (AWS): according to q5 ('Infrastructure Update'), data processing and storage for Business and Enterprise customers; regions EU, US, JP
• ElevenLabs: only for DeepL Voice (substitute voices), according to q1 section 5
• Microsoft Ireland or AWS EMEA: only for DeepL Voice for Meetings (hosting of the bot application, forwarding of audio data to DeepL infrastructure; transcription/translation not on their servers), according to q1 section 5
• Full sub-processor list: only viewable in the Trust Center after an access request ("DPA Sub-Processors", "Get access") (q5/q9); the entries listed here are only those expressly named in q1/q5

• not publicly documented (consumer service, no list)

• Google Cloud sub-processor list: third parties mainly for technical support (e.g. Accenture, Infosys, NTT DATA, TELUS, Virtusa; for all Google Cloud Platform services) and for data labelling in AI services/Pre-Trained APIs (e.g. Cognizant, GlobalLogic, Innodata), each with countries, as well as Google group companies; the list contains no Translation-specific entries (q5)

• Google Cloud sub-processor list: third parties mainly for technical support (e.g. Accenture, Infosys, NTT DATA, TELUS, Virtusa; for all Google Cloud Platform services) and for data labelling in AI services/Pre-Trained APIs (e.g. Cognizant, GlobalLogic, Innodata), each with countries, as well as Google group companies; the list contains no Translation-specific entries (q5)

EUR 0 (q1)

EUR 0 (terms and conditions 1.13: 'provided by DeepL free of charge')

EUR 7.49/month with annual billing, incl. VAT (q3, price for Germany); Write Pro add-on EUR 6.99/user/month (q3)

EUR 24.99/user/month with annual billing, incl. VAT (q3)

EUR 49.99/user/month with annual billing, incl. VAT; 99.0 % availability (q3)

custom (pricing page: 'Custom pricing'); 99.9 % availability (q3)

Pricing page, API tab (German version, retrieved 2026-10-18, q10): Growth plan EUR 23.80/month with annual billing or EUR 29.75 with monthly billing, incl. VAT, plus usage-based costs (12 million characters per year included; EUR 22.00 per 1,000,000 additional characters); Enterprise API: "Individuelle Preisgestaltung" (individual pricing). The pricing page uses the plan names Developer, Growth and Enterprise API.

0 (free service; a paid consumer plan is not documented on the retrieved pages)

In USD (pricing page): NMT text translation and language detection free for the first 500,000 characters per month (as a credit of USD 10 per month, shared between Basic and Advanced, does not roll over, not for formatted document translation); 500,000 to 1 billion characters per month USD 20 per 1 million characters (with Gemini Enterprise Flexible Savings Plan 1 or 3 years: USD 18 or 16); over 1 billion characters per month price via sales. Billed monthly by characters sent. The pricing page contains no information on taxes (q1).

In USD per 1 million characters (pricing page; discounts with Gemini Enterprise Flexible Savings Plan 1/3 years in brackets): NMT text free for the first 500,000 characters per month (USD 10 credit, shared with Basic), then USD 20 (18 / 16); NMT document translation USD 0.08 per page (0.072 / 0.064); custom models text tiered by monthly volume USD 80 up to 250 million, USD 60 up to 2.5 billion, USD 40 up to 4 billion, USD 30 above (with discounts down to USD 24); custom models document translation USD 0.25 per page (0.225 / 0.20); Adaptive Translation USD 25 input and USD 25 output (22.50 / 20); Translation LLM text USD 10 input and USD 10 output; Custom Text Translation (LLM) USD 10 input and USD 10 output (9 / 8). Training custom models USD 45 per hour, at most USD 300 per training job. The pricing page contains no information on taxes (q1).

yes (translation, text improvement; free version)

yes (free API, up to 500,000 characters per month according to the terms and conditions' Service Specification)

yes

yes

yes

yes

yes (pay-per-use or plan; up to 1,000,000 characters per document according to the Service Specification)

yes (machine translation of words, phrases and web pages; Workspace Help: 'Instantly translate words, phrases, and web pages between English and over 100 other languages', q9)

yes (Neural Machine Translation (NMT): language detection and text translation, q1)

yes (NMT model, Translation LLM, Adaptive Translation, custom models (AutoML Translation), batch and document translation (DOCX, PPT, PDF, XLSX), romanisation, glossaries, q1)