iFLYTEK unveils new edge and base large models for enhanced AI capabilities

iFlytek announced plans to open-source two edge-side general large models, Xinghuo X2.5-4B and Xinghuo X2.5-1.7B, on September 1st, 2026. Both models support a context window of up to 1 million tokens and have improved in areas like intelligent agent interaction, mathematical computation, and basic general understanding. These models will support edge-side scenarios such as in-vehicle terminals, smart hardware, and the Internet of Everything. On September 7th, iFlytek will release the Xinghuo X2.5 base model with a scale of 293B, further enhancing code generation and intelligent agent collaboration. During its mid-year performance briefing in 2026, the company mentioned that a new main general large model based on fully domestic computing power will be launched at the Global 1024 Developer Festival later this year, continuing to promote independent innovation in underlying technologies.