Arena AI Model Rankings: Chinese Models Shine, kimi-k3 Tops Frontend Development Chart

The Arena platform released its AI model rankings for Week 30 of 2026. In the overall chart, Anthropic's Claude series dominated the top 6, with claude-fable-5 leading at 1508 points. The top 7 models are closely matched within 30 points. Google's gemini-3.1-pro-preview moved up to 9th. Among Chinese models, Moonshot AI's kimi-k3 ranked 11th (1485 points), the highest; Alibaba's qwen3.7-max-preview and Zhipu's glm series also placed well. In the coding chart, Anthropic's Claude models held the top 5. kimi-k3 rose 2 spots to 8th (1530 points), entering the Top 10. Other Chinese models like Alibaba's qwen3.7-max-preview, Zhipu's glm-5.1, and Xiaomi's mimo-v2.5-pro performed steadily. The frontend development chart saw a breakthrough: kimi-k3 topped at 1682 points, surpassing Anthropic's claude-opus-5-high (1673 points), becoming the world's best in frontend. Zhipu's glm-5.2 (max) ranked 5th, with ByteDance, Tencent, and Alibaba models also in the top tier.