Woofun AI data shows that Kimi K3 achieved an Elo score of 1543 in the Artificial Analysis AA-Briefcase ranking, placing second only to Claude Fable 5 at 1574. The model surpassed GPT-5.6 Sol, Claude Sonnet 5, and Claude Opus 4.8. While K3’s objective pass rate stands at 51% compared to Fable 5’s 56%, its analysis quality score is slightly higher at 1754 versus 1744.

However, K3 underperforms in final product presentation relative to GPT-5.6 Sol and Opus 4.8.

The performance gains involve significantly higher resource consumption. K3 averages $10.57 per task, approximately 10 times the cost of Kimi K2.6. Each task requires an average of 83 rounds, outputs 120,000 tokens, and takes 56.4 minutes to complete. This duration is about 2.5 times longer than that required by Fable 5 for similar tasks.