TORONTO, August 24, 2026 – Thomson Reuters (Nasdaq/TSX: TRI) announced the launch of Thomson, its first proprietary large language model developed in-house. Unlike frontier labs that spend billions on compute and years of infrastructure, Thomson Reuters invested $40 million to train Thomson on a strong open-source foundation, tailored for professional intelligence tasks.

The model leverages decades of proprietary content from Westlaw, Practical Law, Checkpoint, and Reuters, along with input from hundreds of subject matter experts. Early evaluations show Thomson performs on par with the latest frontier models. It is already being used in Co Counsel Legal, with more capabilities planned.

"For years, the AI industry has treated scale as the answer," said Joel Hron, Chief Technology Officer at Thomson Reuters. "Thomson shows there is another path: start with a strong foundation, specialize it deeply, and build intelligence that is highly capable, efficient, and under your control."

Thomson emphasizes AI sovereignty, addressing how models are trained, where they run, and how user privacy is protected. It demonstrates significant improvements in instruction following and navigating dense, domain-specific content—key for complex professional tasks. The model has been trained on less than 10% of Thomson Reuters' content so far, with future development focused on deeper specialization rather than simply adding more data.

"Thomson proves what's possible when you build AI on decades of proprietary content and editorial expertise," said Steve Hasker, CEO of Thomson Reuters.