马斯克频频锐评国产AI模型：二次称赞Kimi K3模型令人印象深刻

Odaily Planet Daily News: Yesterday, Moonshot AI's latest open-source AI model, Kimi K3, was released and ranked first on the Frontend Code Arena testing website with a score of 1679, surpassing Claude's Fable 5 model. Artifacial Analysis conducted relevant evaluations on the K3 model, after which Elon Musk once again praised Moonshot AI's Kimi model, stating that the benchmark performance of the K3 model is impressive.

In March of this year, when Kimi released the research paper "Attention Residuals: Rethinking Deep Direction Aggregation," Musk praised it, saying, "Kimi's research work is impressive." Previously, he also stated that Zhipu AI's GLM model might surpass the Claude Mythos model (i.e., Fable 5) in Q1 2027. In response, Zhipu AI founder Tang Jie said, "It won't take that long."