China’s mysterious AI company Naive AI is valued at over $1.4 billion and could release its first open-source llm model as early as this month.

According to people familiar with the matter, Naive AI, an AI startup founded in February this year by Tsinghua University professor Dai Jifeng, has completed three funding rounds totaling $400 million, bringing its valuation to more than $1.4 billion and earning it unicorn status. Its investors include Tencent, IDG Capital, Matrix Partners China, and Sequoia China. It is understood that Naive AI raised $100 million in its first funding round, $180 million in its second, and $120 million in its recently completed third round, at a post-money valuation of $1.42 billion. Although the company has been around for more than seven months, it has kept a low profile, with fewer than 100 employees. People familiar with the matter said Naive AI will launch its first large language model, also named Naive, as early as this month, releasing it as an open-weight model. In terms of technical approach, Naive AI did not choose the costly option of pretraining from scratch. Instead, it is modifying the architecture of existing china open-weight models and then continuously optimizing them through processes such as mid-training, post-training, and reinforcement learning, in hopes of achieving better performance on multiple tasks. It is not yet possible to confirm which specific base model it uses. In addition, Naive AI is also researching recursive self-improvement technology. That is, the ability of AI models to automatically iterate and upgrade themselves. theinformation.com/articles/tsinghua-pr...llm-startup-hits-1-4-billion-valuation