DeepL
DeepL is a German language AI company founded in Cologne in 2017 by Jarek Kutylowski. He remains its CEO. The company develops translation and writing technology and provides it through applications and APIs.
Company & leadership
- Founder and CEO
- Dr. Jaroslaw (Jarek) Kutylowski.
- Registered office
- Maarweg 165, 50825 Cologne, Germany.
- Legal entity
- DeepL SE; Cologne commercial register HRB 104617.
- Investment
- The May 2024 funding round was led by Index Ventures, with new and existing investors participating.
- Ownership disclosure
- Investor-backed. The cited financing announcement does not disclose a complete current shareholder register or control percentage.
How DeepL began
The project started within Linguee in 2016. Kutylowski led the team that developed the first Translator using neural networks and Linguee's translation data. The public service launched in August 2017 with seven languages. DeepL Pro followed in March 2018 and added paid web use, translation software integrations and an API.
The company's current profile identifies 2017 as its founding year. Its legal imprint identifies DeepL SE as the operator. The product's launch date and the start of research within Linguee describe different stages of this history.
Leadership and investment
Kutylowski leads DeepL as founder and CEO. The company operates from Cologne and has expanded internationally. Its May 2024 announcement reported a $300 million investment at a $2 billion valuation. Index Ventures led the round. ICONIQ Growth and Teachers' Venture Growth joined existing investors including IVP, Atomico and WiL.
The announcement documents a financing round, rather than a sale of the company. It does not provide a complete current shareholder register or controlling ownership percentage.
Expansion into writing and voice
DeepL launched Write Pro in April 2024 using its own proprietary language model technology. This expanded its business from translation into editing and rewriting within a language.
In June 2026, DeepL announced that Mixhalo's team and real-time audio streaming technology would join its platform. The move added audio delivery expertise and established DeepL's first San Francisco office. The announcement links this work to live events, meetings and customer support.
In September 2026, DeepL released new Voice models that preserve a speaker's voice and delivery across languages. It also introduced a desktop application for Zoom, Teams and Google Meet. Voice-to-voice use through that application became generally available; browser use for external participants remained in beta.
Public model scope
DeepL updates its hosted translation models without a selectable checkpoint for each release. Its API selector expresses a quality or latency goal. This directory covers translation, Write and Voice as three families, with language variants and API options grouped within them.
The cited product and API documentation provides hosted access. It does not identify downloadable weights, parameter counts or fixed token windows for these families. Document reconstruction is described as a system that combines VLM extraction, translation and rendering. It is not listed here as a separately named foundation model.