Chinese AI Model Kimi K3 Surpasses GPT-5 in Frontend Code but Fails in Math
Moonshot's Kimi K3, a Chinese AI model, has topped the Code Arena: Frontend rankings, beating OpenAI's Claude Fable 5 and GPT-5.6 Sol. However, it lags far behind in complex math, scoring only 39 percent on FrontierMath Tier 4.
20-Jul-2026
llm
openai
moonshot