If you believe benchmarks, maybe this is true. But I've done my own experiments and it is absolutely not the case for real world usage. The quality of output from Claude (Sonnet) was much higher than Kimi K2.5.