RWS launches AI translation tool to challenge DeepL

RWS has launched Language Weaver Pro, a new artificial intelligence translation model developed with Cohere.

The company says it is designed for businesses and public sector organisations handling business-critical content. Internal benchmarking, according to RWS, placed it ahead of DeepL and Gemini in 31 of the 32 languages tested.

RWS describes Language Weaver Pro as a model with more than 100 billion parameters and says it is the largest dedicated translation model in production. It was built for translation work in which accuracy, fluency and data handling face close scrutiny, particularly in regulated environments and in content carrying commercial or legal risk.

According to RWS, the benchmarking combined human and automated testing. It covered sentence- and paragraph-level datasets, including factual text and marketing material, which can be harder for machine translation systems because of tone, phrasing and cultural references.

The launch comes as translation providers and artificial intelligence companies compete to position their tools for workplace use rather than consumer applications. For many large organisations, translation software is increasingly tied to compliance, internal knowledge systems and customer communications, increasing the importance of security controls and consistency across languages.

Language Weaver Pro is part of RWS's Language Intelligence offering, which combines multilingual artificial intelligence models, translation expertise and secure language technology. It is also integrated into the company's Trados portfolio, used by linguists and localisation teams.

Ben Faes, Chief Executive of RWS, said the product is intended to address a gap in the market.

"Most AI translation speaks the language but misses the meaning. Language Weaver Pro closes that gap - it's the first AI translation solution built to understand not just words, but culture, context and compliance," Faes said.

Cohere, which worked with RWS on the model, focuses on artificial intelligence systems for corporate customers, with an emphasis on data control and deployment options. The partnership reflects a broader trend of software and services groups combining specialist language assets with foundation model providers to build tools for narrower business uses.

Benchmark claims

RWS did not publish detailed scoring data in the launch material, but said Language Weaver Pro ranked first in 31 of 32 languages in its tests. It also said the model delivered higher translation quality than competing tools on both short and longer passages.

This matters because paragraph-level translation can reveal weaknesses that do not appear in isolated sentence tests. Systems may translate individual lines correctly but still lose meaning, register or context when handling connected text, particularly in sectors such as marketing, law, life sciences and government communications.

RWS has long operated in localisation, translation and intellectual property support, and the launch shows its effort to link its linguistic services more closely with artificial intelligence products. The company says its wider language technology draws on decades of linguistic expertise as well as a network of specialists and data professionals.

Enterprise focus

The focus on business and government users also highlights a central commercial question for artificial intelligence vendors: whether customers will pay a premium for tools tailored to specialist tasks. Generic models can often handle translation reasonably well, but organisations in regulated or sensitive sectors may want stronger audit, governance and privacy controls, along with less tolerance for factual or tonal errors.

For Cohere, the partnership is another example of its technology being used in a defined enterprise workflow rather than as a general-purpose chatbot. Its management has argued that corporate adoption will depend not only on model quality but also on whether systems reflect how organisations communicate across departments, markets and legal jurisdictions.

Aidan Gomez, chief executive and co-founder of Cohere, said: "RWS is pushing the boundaries of what's possible in enterprise translation. By building on Cohere's secure frontier AI technology, they can capture meaning, nuance, and context across languages in a way that reflects how people and businesses actually communicate. High-quality translation is essential infrastructure for global businesses operating across borders."