AI News Feed
Market watch
Companies

Thomson Reuters Launches In-House AI Model Built on Alibaba's Qwen

Thomson Reuters launches Thomson-1, an in-house LLM based on Alibaba's Qwen, to cut costs and reliance on closed-source AI.

Thomson Reuters previously integrated Anthropic's Claude into its AI assistant. Citing cost and dependency concerns, the company chose to build Thomson-1 on Alibaba's open-source Qwen3.5 model. The new model will initially handle tasks like table analysis and document review, gradually taking over some work previously performed by Claude.

Open-source models allow enterprises to deploy and customize them independently, offering high cost-effectiveness and security. As Alibaba's Qwen and other open-source models improve, more companies are shifting from closed to open systems. Airbnb CEO Brian Chesky said his company "relies heavily on Alibaba's Qwen models," calling them "better and cheaper than OpenAI." Pinterest CEO Bill Ready said the cost of using Qwen is "less than 8% of comparable closed-source models from Anthropic and OpenAI."

Alibaba has open-sourced more than 460 Qwen models, with total downloads exceeding 3 billion on open-source communities and over 300,000 derivative models, making it the world's largest open-source model family. The recent release of the Qwen3.8 series has sparked a global developer frenzy, with Qwen3.8-27B hailed as "local Opus4.6." Alibaba also said new Qwen3.8 models will be open-sourced this week.