Qwen-Image-3.0-Pro Reaches #5 on LMSYS Text-to-Image Arena Leaderboard
Alibaba's Qwen-Image-3.0-Pro has climbed to the #5 spot on the LMSYS Text-to-Image Arena leaderboard with a score of 1263 points. This represents a significant jump from Qwen-Image-2.0-Pro, which previously held the #15 spot with 1191 points. This update positions Qwen as a highly competitive model in the generative AI space, particularly for practical productivity tasks. The rapid improvement demonstrates the intensifying competition among frontier image generation models. The model saw its largest ranking improvements in the Cartoon category (rising from #16 to #4) and Commercial Design (rising from #16 to #6). Technically, Qwen-Image-3.0-Pro supports native rendering of 12 languages, over 20 fonts, and realistic simulations of digital interfaces like web pages and games.
## BACKGROUND
The LMSYS Text-to-Image Arena is a popular benchmarking platform where users vote on side-by-side image generations from anonymous models to rank their quality. Qwen is a family of AI models developed by Alibaba Cloud, spanning text, vision, and image generation capabilities.