Thomson Reuters launches proprietary LLM trained on Westlaw and Practical Law content
Thomson Reuters has launched its own language model, "Thomson", trained on Westlaw and Practical Law content, challenging Harvey and Legora. According to Startup Fortune it is built on Alibaba's Qwen model at a cost of $40 million, and The New Stack reports that the company nonetheless also uses Anthropic's model.
Who is affected and how
- lawyer
- Legal research tools should be compared on benchmarks and on the firm's own workflows, because the models lead in different areas.
- corporate lawyer
- When switching vendors, the model's origin (a Chinese base model) should also be examined.
Sources
- Thomson Reuters Launches Thomson, Its Own Proprietary LLM Trained on Westlaw and Practical Law Content | LawSiteslawnext.com · Aug 24, 2026
- Thomson Reuters launches proprietary AI model for legal work - SiliconANGLEsiliconangle.com · Aug 24, 2026
- Thomson Reuters launches proprietary legal LLM "Thomson" - Legal IT Insiderlegaltechnology.com · Aug 24, 2026
- Thomson Reuters launches first proprietary LLM to take on legal AI startups - The Global Legal Postgloballegalpost.com · Aug 25, 2026
- Thomson Reuters launches AI model in fresh challenge to Harvey and Legora - The Lawyer | Legal insight, benchmarking data and jobsthelawyer.com · Aug 24, 2026
- Thomson Reuters Launches Proprietary AI Model to Power CoCounsel Legal - LegalTech.calegaltech.ca · Aug 24, 2026
- Thomson Reuters Launches Next-Generation CoCounsel Legal AI Ecosystemprecedenceresearch.com · Aug 24, 2026
- Thomson Reuters trained its own AI model. Then it kept using Anthropic's anyway. - The New Stackthenewstack.io · Aug 24, 2026
Machine-generated summary based on the titles and excerpts of the source articles. The source article is the authority for the facts. News monitoring, not legal advice.