iFLYTEK unveiled the Starfire X2.5 model. The model uses MoE architecture, and the parameter scale is 293B-A30B. It focuses on improving code and intelligence capabilities, and continuously enhances general abilities such as mathematics and understanding. The model is based on nationally produced computing power to complete the whole process of training and reasoning, and continuously optimizes the problem of domestic ultra-long sequence training and promotion. It has now been launched on the iFLYTEK Open Platform.

iFLYTEK unveiled the Starfire X2.5 model. The model uses MoE architecture, and the parameter scale is 293B-A30B. It focuses on improving code and intelligence capabilities, and continuously enhances general abilities such as mathematics and understanding. The model is based on nationally produced computing power to complete the whole process of training and reasoning, and continuously optimizes the problem of domestic ultra-long sequence training and promotion. It has now been launched on the iFLYTEK Open Platform.