Qwen-Image-3.0-Pro debuts at No. 5 on the Text-to-Image Arena leaderboard

Alibaba’s Qwen-Image-3.0-Pro entered the independent Text-to-Image Arena leaderboard at No. 5 with 1,263 points on August 4, 2026, finally giving the image model the benchmark data its July launch lacked.

That is a sharp jump from the prior version: Qwen-Image-2.0-Pro finished 15th at 1,191 points on the same independent leaderboard.

The context is what makes this newsworthy beyond a leaderboard reshuffle. The base Qwen-Image-3.0 model shipped on July 21 with no open weights, no benchmark table and no technical report, leaving buyers little to judge it on. The Arena entry, run by an independent evaluator, supplies the comparison the launch withheld.

Qwen-Image-3.0-Pro is available through an API on Alibaba’s Model Studio and QwenCloud, priced at $0.04 per 1K-resolution output image and $0.075 per 2K-resolution image.

A single leaderboard is a narrow lens. Arena rankings reflect aggregated human preference on its own prompt set, not every production use. Whether Alibaba opens broader access, or publishes the technical detail it skipped in July, will matter more to developers than one benchmark slot.