Thomson Reuters bets $40M on owning its AI instead of renting from OpenAI or Anthropic

With "Thomson," the professional information company is launching its first in-house language model, built on Alibaba's Qwen. The model hits top marks when it can tap into the company's own content and tools.

Thomson Reuters spent about $40 million on staff and computing power over more than two years, according to the company. The more widely touted figure of $450,000 covers only the final training run of the current version. Even the full sum leaves out the real capital: That's decades of content from Westlaw, Practical Law, Checkpoint, and Reuters, plus the working hours of hundreds of domain experts.

The foundation is Alibaba's open Qwen, most recently Qwen3.5-397B, the company says. Working with Imperial College, Thomson Reuters first retrained the Chinese model for safety, ethics, and political neutrality. This intermediate version is called "Snowdon," named after the mountain in Wales.

Then came pre-training on the company's own content, post-training with domain experts, and agentic reinforcement learning inside the company's own tool environments. So far, less than 10 percent of the available content has gone into training.

CTO Joel Hron says the company has "changed the open source starting point like probably close to a half dozen times already." The bigger finding is "less the individual model and more the model factory we built," adds research chief Jonathan Schwartz.

Benchmarks built on a lopsided comparison

The blog post claims Thomson ranks among the best models in the world. The company's own numbers paint a more sober picture: On Stanford LegalBench, Thomson (0.823) trails Gemini 3.1 Pro and GPT-5.5. On the Harvey Legal Agent Benchmark it sits just behind Opus 4.8. It leads on instruction following and the tough PrBench Legal. On reasoning, and especially coding, it falls off sharply. The comparison is also skewed by method, as Thomson competes with test-time scaling, while GPT-5.5 runs without a reasoning mode.

In the  company's in-house Deep Research benchmark with web access alone, Thomson scores 0.53 on factual accuracy, while GPT 5.4 hits 0.65. Only with access to the company's content does Thomson edge past GPT 5.4, 0.83 to 0.82. With web access, Thomson is "within the scope of the other models, but certainly not the leader yet," admits evaluation lead Andrew Bean.

According to Bean, there's "a big uplift that comes from being able to train on and practice with your own tools," something outside providers can't do. What's striking is that GPT 5.4 improves just as sharply with that content. So data access does almost as much work as the specialized training. The company didn't test newer models. The in-house model's razor-thin lead could still grow, though, if the company moves to a stronger variant like Qwen3.8 and pushes far past the 10 percent content mark.

An in-house model instead of frontier fine-tuning

So why build your own model instead of fine-tuning a frontier model from OpenAI or Anthropic on legal data? Thomson Reuters sees three reasons against it. First, the economics: Standard fine-tuning techniques "tend to have a strong tendency to degrade general capability," Schwartz says. And you stay locked into the provider for inference costs and roadmap. A smaller, in-house model pays off precisely on high-volume work like document review.

Second, the data: Training inside its own tools like Westlaw is where the performance jump comes from. The benchmarks prove it, and the company won't hand that access to anyone. Third, the compounding effect, which Hron describes as "renting a house versus buying a house." Every expert review during a product update becomes training data. With an in-house model, "you are building equity in something that you own for the long-term, and that compounds over time," Hron says. With third-party models, that value evaporates at the provider.

For Thomas Reuters the approach works because it can combine three rare ingredients: Exclusive data holdings, hundreds of full-time domain experts, and workflows where quality can be measured objectively. For companies with that profile, the case shows that the open-source community trails the frontier labs by only months, and that $40 million can be enough for a competitive specialized model. Companies that own neither proprietary data nor a way to evaluate results at scale mostly buy ongoing maintenance costs when they build their own model.

According to Hron the next competitive advantage in AI "will come from knowing how to orchestrate it, and knowing which intelligence is important enough to own."

First use in high-volume document review

At launch, Thomson takes over the Tabular Analysis feature in CoCounsel Legal, where a smaller, cheaper model makes economic sense. The product stays multi-model, and administrators can switch. Thomson isn't meant to act as an orchestrator but to handle subtasks like citation checking. Customer data doesn't go into training, the company says.

A small version is coming to Hugging Face as an open-weight model under a non-commercial license, with a technical report and a developer portal to follow. Hron says early, still non-binding talks are underway with law firms about direct licensing.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.