Woofun AI reports that Elon Musk has praised the benchmark performance of Moonshot AI's latest open-source model, Kimi K3. The model achieved a score of 1,679 on the Frontend Code Arena, ranking first and surpassing the Claude Fable 5 model. Musk previously commended Kimi's research paper "Attention Residuals: Rethinking the Aggregation of Depth Direction" in March. He also predicted that the Zhipu GLM model could exceed the Claude Mythos model by Q1 2027, a timeline Zhipu founder Tang Jie suggested would be shorter.