At the recent iFLYTEK Global Developer Festival, iFLYTEK officially launched its new deep reasoning large model - Spark X1.5. This model, based on a fully domestic computing platform, has achieved significant technological breakthroughs and claims to have efficiency exceeding 93% of international competitors, allowing domestic developers to be more confident in the global market.

The highlight of the Spark X1.5 model lies in its breakthrough in the full-chain training efficiency of MoE (Mixture of Experts) models. This means that the model can match international mainstream large models in multiple indicators such as language understanding, text generation, knowledge questions, logical reasoning, mathematical ability, and coding ability. Among them, the mathematical ability still maintains an international leading level, demonstrating iFLYTEK's deep expertise in this field.

In addition, Spark X1.5 also has strong multilingual support capabilities, capable of processing texts in over 130 languages, with overall performance reaching more than 95% of GPT-5. Liu Qingfeng, Chairman of iFLYTEK, stated at the launch event: "Now, Chinese AI developers only need to rely on Spark to go global, without worrying about technical bottlenecks anymore."

The release of Spark X1.5 by iFLYTEK is not only a leap in technology but also provides a "second choice" for the development of China's AI industry, helping to enhance the competitiveness of domestic AI technology in the global market.

With the continuous upgrading of domestic computing power and the ongoing optimization of models, iFLYTEK has demonstrated strong innovation capabilities and market insight. It is bound to shine brightly on the global AI stage in the future.