CoinWorld News, Meituan's new generation foundational large model longcat-2.0-preview has been opened for testing. The total number of parameters in this model exceeds one trillion, ranking among the world's top large models. An insider revealed that the new generation v4 large model released by deepseek on the same day has a total number of parameters and active parameters roughly consistent with Meituan's longcat-2.0-preview. The greater breakthrough of Meituan's new generation foundational large model lies in the fact that its training and inference are entirely reliant on domestic computing power clusters. It is reported that the number of computing power cards used during Meituan's training phase is between 50k and 60k, which is the largest-scale large model training task completed so far using domestic computing power.