Thomson Reuters built its own model on a Chinese open base

Thomson Reuters launched its first proprietary large language model on Monday. The company calls it Thomson. It says it spent $40mn on talent and compute to train the model, and that it began from what its announcement describes as “a…