IT Home AI Model Rankings: Claude-fable-5 Leads, Chinese Models Remain Stable

The latest weekly report from the Arena platform, which benchmarks advanced AI models, indicates that Anthropic's Claude-fable-5 continues to hold the leading position, highlighting the strong presence of U.S. companies in the field. The rankings, covering the period from June 29 to July 5, show a slight point increase for the top model.

In the overall leaderboard, U.S. companies dominate the top positions, with all ten leading models developed by American firms. While minor shifts occurred, such as Gemini-3.5-flash climbing to 11th place, the overall structure at the forefront remained largely unchanged.

Chinese AI models demonstrated stable performance, securing positions in the upper-middle tier. Alibaba Cloud's Qwen3.7-max-preview ranked highest among Chinese models at 15th place. Baidu ERNIE 5.1 and Xiaomi's MIMO-v2.5-pro maintained their spots at 27th and 29th, respectively, indicating consistent capabilities.

In the code generation category, Anthropic models captured nine of the top ten spots. However, Chinese models also showed notable progress. Qwen3.7-max-preview secured the 10th position, while MIMO-v2.5-pro and Baidu ERNIE 5.1 ranked 17th and 29th. This highlights the growing potential of Chinese developers in specialized AI applications.