Beijing based Naive AI raised $400 million across three funding rounds to finance the development of its first AI model. The startup reached a valuation of $1.42 billion with backing from Tencent and other prominent investors. It plans to release its first open weight large language model as early as September by modifying existing Chinese open weight structures instead of pretraining from scratch.

Tsinghua University professor Jifeng Dai founded the firm in February with a staff of fewer than 100 employees. The development process involves modifying the structure of an existing pretrained model and refining it through midtraining, posttraining and reinforcement learning to focus on later stage improvements.