Alibaba's Qwen3.8-Max Ranks 6th in AI Model Blind Test

Arena released its latest blind test rankings for AI models, with several new models performing well. Alibaba's Qwen3.8-Max entered the overall list at No.6 with an ELO score of 1497, tied with the 5th place. Meta's Muse-Spark-1.2 (xHigh) debuted at No.4, becoming a standout new model.

In the coding category, Anthropic's Claude series continues to lead, but Moonshot's Kimi-K3-Max rose to 6th, the highest for a Chinese model. In the front-end development category, Kimi-K3-Max stayed at No.2, Qwen3.8-Max entered the top five, and DeepSeek-V4-Flash-High debuted at No.8. In the agent category, Claude Opus 5 (High) took the top spot, while Kimi K3 (Max) held steady at 5th.

Chinese models have made breakthroughs in several areas, showing growing competitiveness.