DeepL

Taking a moment to look back on a great week at IJCAI-ECAI 2026 in Bremen. It was amazing to be a part of the 35th IJCAI International Joint Conferences on Artificial Intelligence Organization, presenting our paper on Voice translation, engaging in fantastic conversations with fellow researchers, and sharing live demos of the real-time voice layer we’re building at DeepL. Thanks to everyone who stopped by our booth to talk, discuss and explore. And thanks to the amazing DeepL researchers making it all happen: Hans Christian Jung, Johannes Ernesti, Kristina Geißler, Markus Kaiser, Somdeb Majumdar, Max Berrendorf and Leonie Neuhäuser

There’s an inconvenient truth that nobody in AI translation likes to talk about. What we measure in quality test scores is becoming less important than what we don’t. Gaps in MQM-type quality benchmarks may be narrowing, but the differences in real-world performance are becoming much more significant. Our new blog from Morana Perić explains why: https://lnkd.in/eTvQrA94

If you’re using a general-purpose LLM for translation, quality equals latency. If you’re using a specialized model like DeepL, it doesn’t. OpenAI GPT-5.2 is 4x slower than DeepL. Claude Opus 4.6 is 6x slower. Google Gemini 3.1 Pro is a vast 29x slower. If you’re choosing a translation API to build into your products or pipelines, that really matters. Our new blog explains why: https://lnkd.in/ebNwnsv4

Residents and businesses can speak to Walsall Council in the language of their choice thanks to Kerv’s CX Translation real-time voice and digital translation, powered by the DeepL API. Enabling seamless, real-time communication across language barriers ensures callers feel understood, and frees the council from depending on costly third-party interpreters. If you’d like to know more about how Kerv and DeepL enable multilingual customer support, join our webinar with Walsall Council on 17 September at 11am UK time. Register here to book your spot: https://lnkd.in/e38xrDv7

Adding Mixhalo’s ultra-low latency audio to DeepL’s end-to-end language intelligence has transformed what we’re able to build with Voice. When you control the speed of audio delivery, and remove it as a constraint, what was previously impossible quickly becomes a product roadmap. It means speech-to-speech translation, faster than people imagine possible. It means all the excitement of live sport, in any language. And it means that every customer and every patient feel instantly understood. In our latest AI Labs feature, Mixhalo Co-founder Vikram Singh and DeepL Head of Voice Leonardo Doin reveal the real-time audio layer we’re building to transform how the world experiences voice translation: https://lnkd.in/eiENEiNg

A general-purpose LLM can generate a translation for you. What it can’t do is give you translations that are consistent, predictable and controllable, every time. If you’re embedding an AI translation API in products, pipelines or workflows, that really matters. This is Mike Hughes. He leads GTM for our DeepL API and he’s just created a Translation API Buyer's Guide that explains how the model you choose has a big impact on the outputs you get. If you’re choosing a translation API, make sure you read it first: https://lnkd.in/ey274XTw

Cross-border legal work runs on Harvey. The leading legal AI platform’s translations of complex, high-stakes documents now run on DeepL. More than 200,000 lawyers across over 70 countries rely on Harvey to translate contracts, filings, briefs, evidence and more. Thanks to the DeepL API, its clients can upload any document and get a precise, accurate, customized translation. DeepL’s unmatched file format support preserves integrity of structure, meaning and legal terminology. “By integrating DeepL’s AI document translation capabilities directly into the Harvey platform, we’re giving legal teams an even faster, more seamless and reliable way to work across languages, while also making sure content stays highly precise and accurate,” says Harvey’s Product Manager, Lauren Oh. Discover how Harvey works with DeepL: https://lnkd.in/etbepPR3

Choosing a translation API for multilingual products, pipelines or workflows? Here are 8 reasons why you shouldn’t just settle for the first option you find. They’re the dimensions that control translation API performance in practice. And they’re the criteria you should include in your RFP, if you want to ensure quality, latency, price control, developer experience and more. Our new Buyer's Guide has a detailed framework for evaluating translation APIs. You’ll find all the information you need for drawing up a shortlist, creating an RFP and ensuring that the API you choose delivers the performance and pricing you need, at scale. Read the full Buyer’s Guide here: https://lnkd.in/ey274XTw