The Beijing-based startup Naive AI raised $400 million across three funding rounds to attract backing from IDG Capital and HSG. This capital injection pushed the company's valuation to $1.42 billion and added Tencent to its list of investors. Founded in February by Tsinghua University professor Jifeng Dai, the firm currently employs fewer than 100 people and is preparing to launch its first open-weight large language model this month.

Rather than training a model from scratch, Naive AI builds on existing open-weight architectures by adjusting their structure and applying reinforcement learning. The company focuses on midtraining and posttraining to deliver later-stage improvements. This strategy avoids the heavy computational costs associated with full pretraining while iterating on existing Chinese models.