Thomson Reuters launches its own legal LLM

Thomson Reuters launched Thomson, its first proprietary large language model, after investing $40 million in talent and compute and training it on less than 10% of the company’s content, according to the company’s launch announcement. The model is entering CoCounsel Legal, while a smaller open-weight version will be available for academic and non-commercial validation; Thomson Reuters says it controls the model and avoids typical frontier inference costs. Legal-AI buyers should file the move beside the archive’s account of vertical model economics: proprietary domain data can justify owning a model before owning frontier-scale compute.