Chinese AI Unicorn Naive AI Valued at $1.4B Prepares Open-Weight Model Release

Naive AI, a stealth Chinese AI startup founded in February by Tsinghua University professor Dai Jifeng, has reached a valuation of $1.42 billion after raising $400 million across three funding rounds. The company is reportedly preparing to launch its first open-weight large language model, named Naive, as early as this month. The emergence of another heavily funded Chinese AI unicorn committing to open-weight model releases highlights the fierce competition in China's open-source LLM ecosystem. If Naive AI delivers high performance through post-training and architectural modification rather than costly pretraining from scratch, it could validate a more capital-efficient path for open-source AI development. Instead of pretraining models from scratch, Naive AI modifies existing Chinese open-weight architectures and optimizes them using mid-training, post-training, and reinforcement learning. Operating with fewer than 100 employees and backed by prominent investors such as Tencent and Sequoia China, the startup is also researching recursive self-improvement techniques for autonomous model iteration.