Thomson Reuters Unveils 'Thomson,' Its Own Legal-Focused LLM

In one line: Thomson Reuters has introduced Thomson, a large language model built specifically for legal work.

Key points

• Thomson Reuters, the legal-information and professional-services firm, announced a large language model named after itself: Thomson.
• The distinguishing claim is that it is a purpose-built model for the legal domain, rather than a general-purpose model used as-is.
• Detailed specifications — parameter scale, training data, and benchmark performance — should be confirmed against the original announcement.

Why it matters

In fields like law, accounting, and tax, where accuracy and source traceability are critical, demand is shifting toward domain-specific models over general chatbots. Thomson Reuters building its own model reflects a broader "vertical LLM" trend, where firms sitting on proprietary industry data build the models themselves.

Read more

• Meet Thomson: The LLM built by Thomson Reuters — Thomson Reuters Legal Solutions