ModelsAug 25, 2026 · 03:513 sources

Thomson Reuters Builds First Proprietary LLM for $450K to Cut Reliance on Claude

From huggingnews.com

Image: huggingnews.com
AI;DR

Thomson Reuters introduced Thomson-1.0, a proprietary large language model built using Alibaba's Qwen architecture and trained for $450K in compute, at the ILTACON 2026 legal technology conference. The model is designed to reduce reliance on Claude and is initially deployed for Tabular Analysis within the CoCounsel Legal platform, though the service remains multi-model by design.

Read the sourceSign in to suggest

Sources

+ 2 more
  1. SUPPORT@clementdelangue“Thomson Reuters wanted to rely less on Claude”x.com
  2. SUPPORT@sarahcat21“trained for only $450K in compute, can be deployed for a fraction of the cost of frontier models”x.com
#llm#claude#regulationMarkdown
Content history (2)
  • 2026-10-16 · Summary · vi · Replaced stored text
    Tại hội nghị công nghệ pháp lý ILTACON 2026, công ty đã giới thiệu Thomson-1.0, một giải …
    Công ty ra mắt Thomson-1.0 tại hội thảo công nghệ pháp lý ILTACON 2026, cung cấp một lựa …
  • 2026-10-16 · Title · vi · Replaced stored text
    Thomson Reuters chi 450.000 USD xây dựng LLM riêng nhằm giảm phụ thuộc vào Claude
    Thomson Reuters xây dựng LLM đầu tiên nội bộ chỉ với 450K USD để giảm phụ thuộc vào Claude

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today