Alibaba's Qwen image editing model for instruction-based image modifications and transformations
Settled by community votes across 1 shared challenge, with an AI judge weighing in on each.
Qwen Image Edit 2511
19.9
arena score
#15 of 23 in Image Editing
Skill signature
Not enough comparable category data
The chart appears once both models have ratings across at least three shared arena categories.
Z-Image Turbo
24.7
arena score
#15 of 44 in Text-to-Image
Vote tally
Where the votes landed
Qwen Image Edit 2511
100.0%
win rate
Ties
0.0%
Z-Image Turbo
0.0%
win rate
100.0%
0.0% ties
0.0%
Shared challenges
1
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Bald man challenge
Image EditingEdit instruction
“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
Qwen Image Edit 2511
Z-Image Turbo
100% wins
0% ties
0% wins
AI Judge Analysis
Qwen Image Edit 2511
- + Successfully added a full head of hair with realistic texture.
- + Excellent preservation of the original facial features and environment.
- + Natural-looking hairline and seamless blending with the face.
- − None observed; the hair matches the character's aesthetic well.
Z-Image Turbo
- + Preserved the overall composition and lighting of the scene.
- − Failed to add a full head of hair, only adding a very thin layer of fuzz.
- − Significantly altered the background vegetation unnecessarily.
- − Did not meet the 'thick' and 'full' requirement of the prompt.
Verdict: Qwen Image Edit 2511 followed the instructions perfectly, providing the subject with realistic, thick, curly hair that preserves the original person's identity and lighting. Z-Image Turbo largely failed the request, only adding a slight buzz-cut texture while also changing the layout of the background bushes. Qwen Image Edit 2511 is the clear winner for its high-quality, targeted edit.
Next steps
Explore each model
Tongyi-MAI's 6-billion parameter distilled text-to-image model optimized for speed, achieving high-quality generation in 8 steps or fewer with support for bilingual text rendering