Musk noted that the new model could outperform Kimi in quality while maintaining the high generation speed and computational efficiency characteristic of Grok 4.5.
The discussion was sparked by a comparison published by researcher Ming Choi. He stated that Grok 4.5 uses about 1.5 trillion parameters, while Kimi K3 uses 2.8 trillion. Moreover, according to him, running tasks on Kimi costs about three times more. The researcher called this result evidence of the high efficiency of xAI's architecture.
Earlier, Chinese company Moonshot AI introduced a new artificial intelligence model, Kimi K3, which the developer claims is the largest open-source AI model in the world.
Elon Musk reported that the Grok 4.5 language model took first place in the Long-Horizon Terminal-Bench ranking for July 2026.