Thomson Reuters announced on August 24, 2026, the launch of "Thomson," a proprietary Large Language Model (LLM) developed to meet the specific demands of professional-grade work. The move addresses the limitation where general-purpose AI often fails to perform specialized tasks effectively even when provided with organizational data.
Model ReleasesThomson ReutersThomson
Thomson Reuters Unveils "Thomson," a Proprietary LLM for Specialized Professional Services
Built on the Qwen 3.6-35B-A3B architecture, "Thomson" is designed to integrate complex, professional-level information into existing workflows. The model is intended to enhance a suite of services, including Westlaw, Practical Law, Checkpoint, and Reuters.
Thomson Reuters stated that the model aims to achieve "fiduciary-grade" reliability, ensuring the accuracy and specialized knowledge required by professional users in high-stakes environments.
Sources
- AIを導入しても“組織的には使われない”のはなぜ？ 「個人利用」にとどまる理由と脱却策 （ITmedia AI+, 2026-09-24）
- 「汎用AIにデータを渡すだけ」では専門業務をこなせない？ トムソン・ロイターが独自LLM「Thomson」を構築した理由 （ITmedia AI+, 2026-09-24）