AI Large Model Weekly List: Domestic models break through in niche areas, Alibaba wan3.0 ranks top three in video category
2026-09-07 16:34Models🔥 47.3 heat score
2sources
1days unfolding
47.3heat score
10mentions
SummaryAI generated
On September 7, the Arena platform released the rankings for the 36th week of AI large-scale model blind tests. In the comprehensive list, Claude-fable-5 continued to lead with 1507 points, and the top position was still dominated by overseas models. In specific areas, domestic models performed significantly well: Alibaba’s Qwen3.8 series entered the top 15 in the front-end development category for the first time, with Qwen3.8-max-0902 ranking 4th; Tencent’s Hy4-preview and Zhipu GLM-5.3-flash also made it into the top 15. In the video category, Alibaba’s Wan3.0 followed closely behind Google’s Gemini-omni series with 1494 points, placing it among the top three. The top three in the code category were still Anthropic models: Moon’s Dark Side Kimi-k3-max ranked 12th in the comprehensive list, Zhipu GLM-5.3-max ranked 20th, and Alibaba’s Qwen3.8-max ranked 22nd.
According to IT Home's report on September 7, the Arena platform released the 36th week’s AI large model blind testing rankings. Alibaba’s wan3.0 made its debut and quickly ranked among the top three in the video category. This edition of the rankings saw several new domestic models make it onto the list: Alibaba’s qwen3.8-max-0902 placed 4th in the front-end development category, while Tencent’s hy4-preview and Zhipu’s glm-5.3-flash also ranked within the top 15. In the AI video category, Alibaba’s wan3.0 finished second with 1494 points, behind Google’s gemini-omni series. The top positions in the overall rankings were still held by overseas models, with Claude-fable-5 continuing to lead. Domestic models generally ranked in the middle to upper ranks: kimi-k3-max came in 12th, Zhipu’s glm-5.3-max ranked 20th, and Alibaba’s qwen3.8-max was 22nd. The top position in the code category was held by Anthropic’s model, while Google’s gemini-3.8-flas…
According to a report by IT Home on September 7, the Arena (arena.ai) platform released the rankings of AI large models for the 36th week. Alibaba’s qwen3.8-max-0902 ranked 4th in the front-end development category for the first time, while Alibaba’s wan3.0 entered the top three in the video category for the first time. In the overall rankings, Claude-fable-5 remained at the top with 1507 points; the top three in the code category were all Anthropic models, and domestic model glm-5.3-flash rose to 9th place. In the front-end development category, besides qwen3.8-max-0902, qwen3.8-flash-next, Tencent’s hy4-preview, and Zhipu’s glm-5.3-flash all made their debut and entered the top 15. Overall, the leading positions were still held by overseas models, but domestic models have made significant progress in specific areas.