A US artificial intelligence start-up has announced the development of its first proprietary AI model built upon a Chinese open-weight system, signaling a notable shift in Western tech approaches amid rising development expenses. Harvey, a San Francisco-based legal technology company backed by prominent investors including OpenAI, Sequoia Capital, and Andreessen Horowitz, unveiled its new model, Harvey Tenet, on Thursday. The company revealed that this model was post-trained using Moonshot AI’s Kimi K3, an open-weight model developed by the Chinese lab.
Harvey said Harvey Tenet delivered "state-of-the-art" results on complex legal tasks, a key capability for the start-up that serves major international law firms and enterprise clients. Previously, Harvey's AI solutions were built by customizing closed proprietary models from leading US providers such as Anthropic, OpenAI, and Google. The adoption of an open-weight model from a Chinese laboratory represents a marked departure from this earlier strategy.
The move reflects a broader trend within the AI industry where Western companies increasingly leverage open-weight models. These models allow developers to refine or "post-train" general base systems with specialized data to enhance performance on targeted tasks, often resulting in improved accuracy and reduced inference costs. AI policy researcher Simon Hedlin noted on social media that Harvey’s approach exemplifies the benefits of open-weight frameworks, which can be adapted for industry-specific applications at relatively lower cost.
Hedlin also commented on the broader competitive landscape, suggesting that the United States is currently behind in developing advanced open-weight AI models, a situation he described as "unfortunate." He emphasized that the full potential of these highly capable open-weight models remains largely unexplored.
The use of Chinese-open weight models by a US start-up like Harvey illustrates the complex global interplay in AI development and raises questions about future innovation dynamics as companies navigate rising costs and technological challenges in the rapidly evolving AI sector.