Reportedly: iFlytek releases Spark X2.6 end-cloud unified large model, first to achieve code and agent capability breakthrough on fully domestic compute
Reported by IT Home · not yet confirmed by the company or a second independent outlet. We update this page when it is.
iFlytek released Spark X2.6, an end-cloud unified large model, at its 1024 Developer Festival, IT Home reported. The company claims it is the first to achieve code and agent capability breakthroughs using fully domestic compute. Chairman Liu Qingfeng said only Spark is trained on domestic compute, while others use Nvidia cards, and iFlytek's model is the smallest.
China context
- Original name
- 科大讯飞
- Outside China
- Not stated in the sources yet
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers outside China cannot verify the domestic compute claim or access the model unless iFlytek provides international API or weights.
- For investors
- The claim of fully domestic training may signal reduced exposure to export controls for iFlytek, but independent verification is needed.
Translated from Chinese. Quotes and facts link to the original sources.
At the iFlytek Global 1024 Developer Festival, iFlytek officially released Spark X2.6, an end-cloud unified large model. The company claims it is the first to achieve breakthroughs in code and agent capabilities based entirely on domestic compute. Chairman Liu Qingfeng stated that only Spark is trained on domestic compute, while all others are trained on Nvidia cards, and iFlytek's model size is the smallest.
The model is described as end-cloud unified, suggesting a single architecture deployed across cloud and edge devices. The claim of training on fully domestic compute implies use of Chinese-made chips, but no specific hardware or training details are provided. The model size is stated to be the smallest among competitors, but no parameter count is given.
Chinese AI developers gain a model that reduces reliance on Nvidia GPUs, potentially lowering compute costs and easing export-control constraints. Competitors using Nvidia cards may face pressure to demonstrate domestic-compute training to align with policy goals.
For enterprises in China, Spark X2.6 may offer a compliant option for AI deployment on domestic infrastructure. Outside China, availability and performance remain unverified.
Verification of the domestic compute claim and independent benchmarks for code and agent capabilities are needed. Watch for iFlytek to release technical details or third-party evaluations of Spark X2.6.