[Elon Musk Frequently Comments on Domestic AI Models: Praises Kimi K3 Model for the Second Time as Impressive] Odaily Planet Daily News – Yesterday, the latest open-source AI model from Dark Side of the Moon, the Kimi K3 model, was released. It ranked first on the Frontend Code Arena testing site with a score of 1679, surpassing Claude's Fable 5 model. Regarding the K3 model, Artifacial Analysis conducted a related evaluation, after which Elon Musk once again praised the Dark Side of the Moon's Kimi model, stating that the benchmark performance of the K3 model is impressive. In March of this year, when Kimi published the research paper *Attention Residuals: Rethinking Depthwise Aggregation*, it also received Musk's commendation: 'Kimi's research work is impressive.' Previously, Musk had also remarked that the GLM model under Zhipu could potentially surpass Claude's Mythos model (i.e., Fable 5) by Q1 2027. In response, Zhipu founder Tang Jie stated, 'It won't take that long.'