1 AI Model to Rely Less on Anthropic
Thomson Reuters launched its AI model, Thomson-1, based on Alibaba's open-source Qwen model. The company aims to reduce reliance on costly AI from Anthropic, focusing on document review. Thomson Reuters will still use Anthropic's Claude for its AI legal assistant, CoCounsel. The shift is driven by cost savings and leveraging in-house expertise, according to CTO Joel Hron.
How this was made
The 30-second read
Why it matters
The launch of Thomson‑1 aims to cut AI licensing costs and create proprietary capabilities for document review.
Market read
The move could reshape AI sourcing strategies for large data firms and influence the competitive landscape of AI service providers.
What to watch
Integration challenges, data privacy concerns, and potential regulatory scrutiny of Chinese‑origin models.
Background
Thomson Reuters, a major provider of legal and financial data, has historically relied on external AI services like Anthropic's Claude.
Ticker impact
Thomson Reuters launched its own AI model Thomson-1, reducing reliance on Anthropic's Claude.
Potential modest upside if cost savings are realized; limited immediate price move.
No financial figures disclosed; impact depends on adoption and performance of Thomson‑1.
Market effects
Signals a shift toward in‑house AI solutions, pressuring third‑party AI vendors.
May affect North American legal‑data and information‑services providers.
Highlights growing use of Chinese open‑source models in Western enterprises.
Counterpoint
Building a proprietary model may lag behind leading AI platforms and could increase R&D risk.
Key entities
- companyThomson Reuters
Global data and legal information provider (NYSE: TRI).
- companyAnthropic
AI startup behind Claude, current AI partner of Thomson Reuters.
- companyAlibaba
Chinese tech conglomeror that developed the Qwen open‑source model.

