Naive AI Reaches $1.4B Valuation, Open-Weight LLM Launch Imminent

Naive AI, an AI startup founded in February by Tsinghua University Professor Dai Jifeng, recently completed its Series C funding round, raising a total of $400 million and pushing its valuation above $1.4 billion to join the ranks of unicorns.

Major investors including Tencent, IDG Capital, Matrix Partners China, and Sequoia China participated in this investment. The amounts raised per round were $100 million in Series A, $180 million in Series B, and $120 million in Series C, with the post-Series C valuation assessed at $1.42 billion.

Model Release and Technical Strategy

Naive AI plans to release its first large language model (LLM), 'Naive', as an open-weight model as early as this month, just seven months after its founding. The company currently maintains a small workforce of fewer than 100 employees.

Technically, instead of investing heavily in pretraining, the company adopted a strategy of modifying the architecture of existing Chinese open-weight models and optimizing performance through mid-training, post-training, and reinforcement learning. The specific base model has not yet been confirmed.

Research on Self-Improvement Technology

The company is also researching recursive self-improvement technology, where AI models iteratively upgrade themselves.