iFlytek Launches Spark X2-Flash Large Language Model with MoE Architecture
iFlytek has officially launched Spark X2-Flash and opened its API. The model utilizes an MoE architecture with 30B total parameters, supports a 256K ultra-long context, and was trained on a domestic computing cluster based on Huawei Ascend 910B. It features significant improvements in agent and coding capabilities, with actual tests showing performance close to trillion-parameter models, while Token usage costs are less than one-third of mainstream large models. The model has been integrated into platforms such as AstronClaw and Loomy, and is compatible with mainstream Agent frameworks like OpenClaw.
iFlytek has officially launched Spark X2-Flash and opened its API. The model utilizes an MoE architecture with 30B total parameters, supports a 256K ultra-long context, and was trained on a domestic computing cluster based on Huawei Ascend 910B. It features significant improvements in agent and coding capabilities, with actual tests showing performance close to trillion-parameter models, while Token usage costs are less than one-third of mainstream large models. The model has been integrated into platforms such as AstronClaw and Loomy, and is compatible with mainstream Agent frameworks like OpenClaw.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.