iFlytek has launched its Spark X2 large model, which is trained entirely on domestic computing resources. The upgrade features two key advancements: first, a comprehensive enhancement in general capabilities, with Spark X2 now matching top international models in areas such as mathematics, reasoning, language understanding, and intelligent agents. Its multilingual capabilities, covering over 130 languages, continue to improve, maintaining leading performance in key languages for regions like Latin America and ASEAN. Second, leveraging algorithmic breakthroughs, high-quality industry datasets, and reinforced learning with professional reasoning chains, the Spark industry model has been upgraded, resulting in significant progress in product application solutions.

iFlytek stated that this practical upgrade, aimed at key industries, marks a new stage in algorithmic and engineering innovation for China's large model infrastructure built on fully domestic computing power. While achieving overall improvements in general capabilities, Spark X2 places greater emphasis on high-expertise, high-experience scenarios such as education, healthcare, automotive, and intelligent agents, striving to provide leading and reliable technological solutions and support for these fields.

The latest capabilities of iFlytek Spark X2 are now accessible via the Spark web version and mobile app, with a new API also available on the iFlytek Open Platform.