Thomson Reuters Launches "Thomson" — First Proprietary Legal LLM Inside CoCounsel

Thomson Reuters launched Thomson, its first proprietary large language model, on August 24, built by combining the company's legal content archive with frontier model infrastructure. Thomson debuts inside CoCounsel's Tabular Analysis capability — a high-volume document review function — and becomes the default model there while CoCounsel remains multimodel for tasks where frontier models hold an edge. The move signals that Thomson Reuters is no longer purely a model-agnostic integrator; it is building proprietary AI assets that compound over time as more legal content flows through the system. For competitors and adjacent vendors, this raises the question of what "legal domain expertise" means when the dominant content provider begins encoding that expertise at the model layer — not just as retrieval context but as trained capability.