OpenAI-backed legal tech firm pivots to Chinese Kimi K3 open-weight model
Harvey, a US legal artificial intelligence startup backed by OpenAI, has developed its first in-house model, Harvey Tenet, using China's Moonshot AI’s Kimi K3 open-weight base. This move reflects a growing trend among Western tech firms to utilize Chinese open-weight systems due to rising development costs. Harvey Tenet, post-trained on comprehensive legal datasets, reportedly achieves state-of-the-art performance in complex legal tasks, outperforming its underlying base model and US frontier systems like Fable 5 and GPT-5.6 Sol. The company, founded in 2022, reached an US$11 billion valuation in March and serves major international law firms. This pivot from customizing closed proprietary models by US leaders such as Anthropic, OpenAI, and Google marks a significant shift for the firm.
The shift by US legal tech firm Harvey to build its new model, Harvey Tenet, on China’s Moonshot AI Kimi K3 open-weight base highlights the increasing viability of Asian open-weight models for specialized applications. This adoption by an OpenAI-backed company with an US$11 billion valuation suggests that cost efficiency and performance in niche domains are driving Western firms to look beyond traditional US proprietary systems. The use of Kimi K3, post-trained with legal data, demonstrates that Chinese foundational models can be competitive for specific enterprise needs, even if broader adoption by US firms like AT&T remains cautious due to risk evaluation. For Asian markets, this development points to a potential acceleration in the development and adoption of local open-weight models. As more Western companies validate the performance and cost benefits of systems like Kimi K3, it could encourage further investment and innovation within the Asian AI ecosystem. The ability to achieve "state-of-the-art" performance in complex legal work with an Asian base model offers a strong case for regional developers to explore and refine their own open-weight offerings, potentially reducing reliance on US-centric AI infrastructure. The thing to watch is how this trend impacts the competitive landscape for AI talent and investment in Asia. If Chinese open-weight models gain wider international traction, it could attract more global capital and expertise to Asian AI labs, fostering a more robust and diverse AI development environment. This move by Harvey, a company that previously relied on US closed models, suggests a pragmatic, performance-driven approach that could benefit Asian AI innovation.
