RWS introduces Language Weaver Pro.

      The language technology company listed on AIM claims that its new model, which is the largest dedicated translation model currently in use, achieved the top rank in 31 out of 32 languages during internal tests against DeepL and Gemini. This model is fully integrated into the Trados portfolio.

      RWS highlights a genuine issue in enterprise translation: the fastest tools, which are general-purpose AI models that also provide translation, often deliver output that is fluent but contextually inaccurate, inconsistent across different uses, and primarily designed for consumer needs rather than for regulated sectors.

      For instance, a legal contract, a pharmaceutical label, or a government policy document cannot tolerate ambiguity in the same way a marketing email can. In response to this challenge, RWS has introduced Language Weaver Pro, a new enterprise translation model aimed at addressing these specific needs.

      Language Weaver Pro boasts over 100 billion parameters and was developed in collaboration with Cohere, a Canadian enterprise AI firm established in 2019 by Aidan Gomez.

      RWS claims it is the largest dedicated translation model currently operational and asserts that internal benchmarking—utilizing both human-led and automated evaluations across sentence and paragraph-level datasets comprising factual and marketing content—placed Language Weaver Pro first in 31 out of 32 languages compared to DeepL and Google Gemini.

      The model is now available and seamlessly integrated into the Trados portfolio, RWS’s translation management platform, allowing enterprise teams to utilize it within their existing localization workflows.

      The exclusive partnership with Cohere combines two complementary strengths: Cohere's security-focused AI model infrastructure and RWS’s extensive linguistic fine-tuning data, domain expertise, and experience in enterprise integration.

      This collaboration is branded as RWS’s Language Intelligence capability, which represents a more comprehensive positioning that includes the new model’s infrastructure along with the governance and compliance tools needed by enterprise and government clients.

      Gomez, CEO and co-founder of Cohere, presented it as an essential infrastructure component: “High-quality translation is vital infrastructure for global businesses operating across borders.”

      RWS’s competitive stance against DeepL and Gemini is the most direct and striking aspect of this announcement. DeepL has established a strong enterprise translation business by asserting that dedicated translation models surpass general-purpose AI.

      RWS essentially echoes this sentiment while challenging both DeepL and the general-purpose incumbents at once, asserting that its larger model is trained on more specialized data.

      Founded in 1958 and listed on AIM (RWS.L) since 1997, RWS employs approximately 250,000 data specialists, cultural experts, and language professionals across its network and holds over 45 AI-related patents. The company caters to more than 80 of the world’s top 100 brands, including the majority of leading pharmaceutical companies and patent applicants.

      Ben Faes, who took on the role of CEO in January 2025 after managing Google Cloud for Southern Europe and Emerging Markets, is steering the company's shift from traditional translation services to AI-centered solutions. Language Weaver Pro represents the most significant implementation of this strategy to date.