Harvey Launches Legal AI Model Tenet with Significant Performance Improvements

Legal AI startup Harvey has launched its first post-training model for legal scenarios, named Tenet. Built on Moonshot's Kimi K3, Tenet was developed through post-training by Harvey and Fireworks AI using publicly available legal data, synthetic data, and lawyer-annotated data. In the LAB test set, Tenet's all-pass rate improved by 82% compared to the base model, with a 22% increase on LAB Contracts, where it secured the top position, while LAB ranked second. Harvey has also trained three sub-models around Tenet, targeting M&A due diligence, high-frequency document review, and structured extraction, with inference costs less than a quarter of mainstream foundational models. Investor David Sacks noted that U.S. startups can create high-performance industry models at low costs through open-source bases and vertical post-training paths. If overly strict restrictions are placed on open-source models, it will be difficult to prevent Chinese labs from launching the next Kimi, but it will weaken the ability of startups like Harvey to build models in specialized fields.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.