AI Model Rankings: Alibaba's Qwen3.8-Max Climbs to Global Top 6

The Arena platform released its weekly language model rankings for the week of August 3-9, 2026. Multiple new models entered the competition, with Alibaba's Qwen3.8-Max making a strong showing, claiming sixth place in the overall rankings with 1497 ELO points and establishing itself among global leaders. Meta's Muse Spark 1.2 (xHigh) secured fourth position with 1498 points, marking the emergence of new competitors. At the top, Anthropic's Claude Fable 5 maintains its lead with 1507 points, closely followed by Claude Opus 4.6 Thinking and Claude Opus 4.7 Thinking with minimal differences. Chinese models like Moonshot's Kimi K3 Max continue to advance, particularly in specialized rankings. In the image generation category, Alibaba's Qwen-Image 3.0 Pro and ByteDance's Seedream 5.0 Pro secured positions in the top 10. In the video rankings, MiniMax's Mini-Max H3 made its debut at fourth place, solidifying Chinese model presence among the most capable systems.