Thomson Reuters Invests $40M to Train a Fully Controlled Frontier AI Model
Thomson Reuters invested $40 million to train its own frontier model using an open-source foundation plus decades of proprietary content and domain expertise. The model is fully controlled by Thomson Reuters and aims to reduce inference costs versus typical frontier systems, targeting “Fiduciary-Grade” standards and early evaluations that match leading frontier models across tasks.
Coverage (4)
- Thomson Reuters launches in-house AI model to cut Anthropic costs (finance.yahoo.com)
- house AI model to cut Anthropic costs (qz.com)
- Thomson Reuters Leverages its World-Class Data Assets to Launch Its Own Frontier Model (prnewswire.com)
- Thomson Reuters Leverages its World-Class Data Assets to Launch Its Own Frontier Model (morningstar.com)