RWS has launched Language Weaver Pro, a new enterprise AI translation system the company says outperformed DeepL and Gemini in benchmarking across 31 of 32 languages.
The company said the product is designed for business-critical content where accuracy, fluency, security, and governance matter more than the speed-first model of consumer-grade tools. Built in partnership with Cohere, Language Weaver Pro is based on a 100+ billion parameter model, which RWS describes as the largest dedicated translation model in production.
Ben Faes, CEO of RWS, said: “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.”
RWS said human-led and automated benchmarking showed stronger performance than rival systems across both sentence- and paragraph-level datasets, covering factual material as well as more demanding marketing content. The launch sits within a wider Language Intelligence proposition that combines multilingual AI models, linguistic expertise, and secure language technology for regulated and complex environments.
Aidan Gomez, CEO 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.”
RWS added that Language Weaver Pro is natively integrated across its Trados portfolio, giving linguists and organisations access to the model inside tools already used throughout the localisation process. That is significant for enterprises looking to deploy AI translation inside existing workflows rather than through standalone consumer applications.
The product is available immediately, positioning the launch as part of a broader push to make AI translation core infrastructure for regulated, multilingual, and globally distributed organisations.
