Collection
Thomson Reuters launched an LLM trained on 175 years of its own data
Thomson Reuters launched Thomson, an LLM that starts from Qwen3.5-397B and is trained on 175 years of the company's proprietary legal, tax, accounting, and news data. Zach Moskow said the company spent $40 million on it. Thomson Reuters' published benchmarks put the model near Claude Opus 4.8; those scores are the company's claim, not an independent test. Also in this pack: Kalpa TTS Beta v0.1 is listed No. 2 on Design Arena's Audio Realism Bench at Elo 1331, ahead of MAI-Voice-2 and Eleven v3 C
Thomson Reuters launched an LLM trained on 175 years of its own data
Open originalZach Moskow reported that Thomson Reuters launched Thomson, an LLM that starts with Qwen3.5-397B and is continually trained on 175 years of proprietary legal, tax, accounting, and news data. He said the company spent $40 million, and that Safe Sign, a legal-AI startup it bought two years ago, built the model. Thomson Reuters' published benchmarks claim it is comparable to Claude Opus 4.8 and ahead of GPT-5.5, Claude Sonnet 5, and Gemini 3.1 Pro overall. Those rankings are the company's claim, not an independent evaluation.
Kalpa TTS Beta v0.1 ranks No. 2 on Audio Realism Bench
Open originalDesign Arena said Kalpa TTS Beta v0.1 by Kalpa Labs took No. 2 on Audio Realism Bench with an Elo of 1331. The post says it ranks ahead of Microsoft's MAI-Voice-2 and ElevenLabs' Eleven v3 Conversational, and behind Bland Speech v3. The Elo and that order are Design Arena's reported result, not an independent audit.
Lauren published part 1 of a pstack guide
Open originalLauren (@poteto) published part 1 of a written guide to pstack as an X article titled The Complete Guide to pstack Pt. 1. The post is the announcement of that article, not a product launch.