Kimi K3 Draws Investor Attention After Strong Coding Benchmark Results

CITIC Securities said Moonshot AI’s Kimi K3 has drawn broad investor attention after its strong performance, with Arena ranking it first on a code benchmark and making it the first open-source model to match top global closed models in a blind test in coding. According to Jiemian News, overseas investors have called it another “DeepSeek moment” for China. The research note said K3 stood out for its performance on multiple benchmark tests and for sharply lower token-level calling costs. However, some users later reported that, due in part to limited computing power, its end-to-end reasoning time can be relatively long on complex tasks, and its token usage remains high, leaving task-level costs a challenge. CITIC Securities said Kimi K3 shows Moonshot AI’s continuing technical accumulation and competitiveness, and argued that Chinese model makers led by Kimi are iterating under constrained computing resources to narrow the performance gap with top U.S. models.