OpenAI
1306 Elo ±45
Rank #2 of 42 · median 126s per image · 12 refused
1–0–0
wins – ties – losses for GPT Image 2
1 direct comparisons
Elo expectation: GPT Image 2 wins 60%
Alibaba
1236 Elo ±57
Rank #9 of 42 · median 77s per image

By category

CategoryGPT Image 2Qwen-Image-3.0-ProDirect record
Anime & Cartoon Style 1344 1207 —
Architecture & Interiors 1531 1415 —
Complex Scenes 1443 1326 —
Ghibli style 1195 1270 —
Graphic Design 1208 1180 —
Hands & Anatomy 1128 1213 —
Photorealistic People & Portraits 1349 1127 1–0–0
Surreal & Creative Prompts 1366 1234 —
Text in Images 1356 1318 —
Ultra Hard 1334 1310 —

By criterion

CriterionGPT Image 2Qwen-Image-3.0-Pro
Prompt adherence 1219 1157
Technical quality 1235 1228
Artistic merit 1213 1197
Detail execution 1276 1230
Realism & coherence 1239 1185

Side by side

A 90s anime-style space battle scene with giant starships and laser beams. - GPT Image 2
GPT Image 2 9 / 10 An excellent, highly convincing retro-anime interpretation with strong scale and narrative presentation. Slightly reduced background readability keeps it behind the two leading images, rather than indicating any major failure.
A 90s anime-style space battle scene with giant starships and laser beams. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 7 / 10 Ranked 5th of 13, this is imaginative and richly executed, with particularly effective retro missile choreography. Uneven mechanical perspective and crowded effects prevent it from matching the clearer top-tier scenes.
A traditional Moroccan riad courtyard with detailed zellige tilework, flowing fountain, citrus trees, and ornate woodwork. Morning light casting intricate shadow patterns. - GPT Image 2
GPT Image 2 9 / 10 Rank 3 of 13: an excellent combination of cultural specificity, naturalistic materials, and inviting composition. It belongs to the leading group, with only minor fine-detail limitations.
A traditional Moroccan riad courtyard with detailed zellige tilework, flowing fountain, citrus trees, and ornate woodwork. Morning light casting intricate shadow patterns. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 7 / 10 Ranks sixth of 13 for its coherent, restrained, and believable interpretation. It lacks the decorative precision and especially compelling light patterns that distinguish the highest-ranked courtyards.
An astronaut and a deep-sea diver playing chess together inside a submarine. - GPT Image 2
GPT Image 2 8 / 10 This ranks sixth, offering excellent finish and strong character interaction. The contradictory environmental readings and slightly congested chess details reduce its coherence relative to the higher-ranked entries.
An astronaut and a deep-sea diver playing chess together inside a submarine. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 4 / 10 Ranked 13th of 13, the apparent third astronaut hand is the clearest critical anatomical defect in the set. Successful submarine detailing cannot compensate for that failure in a primary subject.
A child and elder tending oversized vegetables in a garden. Reproduce the artistic style of Studio Ghibli, with their distinctive character proportions and detailed plant illustrations. - GPT Image 2
GPT Image 2 8 / 10 Ranks fifth, offering strong stylistic fidelity and believable teamwork. Its subdued contrast and crowded vegetation make it slightly less immediately compelling than the higher-ranked scenes.
A child and elder tending oversized vegetables in a garden. Reproduce the artistic style of Studio Ghibli, with their distinctive character proportions and detailed plant illustrations. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 6 / 10 Ranked 10th of 13: technically accomplished detail and clear gardening activity are offset by a significant stylistic mismatch. The hybrid realistic-and-anime treatment makes it less successful for this particular prompt than more consistently illustrated alternatives.
Abstract background graphic, cyberpunk aesthetic, neon pink/blue light trails on dark grid, sense of digital speed, for desktop wallpaper. - GPT Image 2
GPT Image 2 9 / 10 Ranked 1st of 13 for the strongest combination of abstraction, forward momentum, technical finish, and distinctive composition. Minor signage and considerable visual density prevent a flawless score.
Abstract background graphic, cyberpunk aesthetic, neon pink/blue light trails on dark grid, sense of digital speed, for desktop wallpaper. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 6 / 10 Ranked 8th of 13, this is a competent abstraction with useful negative space. Less controlled intersections and uneven secondary detail prevent it from matching the stronger compositions.
A hand drawing a sketch on paper with a pencil, the hand and pencil tip in clear focus. - GPT Image 2
GPT Image 2 5 / 10 The illustration and photographic surfaces are accomplished, but the apparent extra lower digit is a critical weakness for this anatomy-centered task. That flaw places it below simpler images with clearly believable finger structure.
A hand drawing a sketch on paper with a pencil, the hand and pencil tip in clear focus. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Ranked 1st of 13 for the strongest combination of realistic posture, precise contact, and tactile materials. Its restrained botanical composition and atmospheric lighting give it a slight advantage over the other excellent entries.
A ultra high-resolution portrait of an elderly woman with silver hair and bifocal glasses, each lens reflecting a distant landscape, captured in warm late-afternoon sunlight - GPT Image 2
GPT Image 2 9 / 10 Rank 1 of 13: this provides the strongest integration of realistic aging, warm light, landscape reflections, and identifiable bifocal cues. Minor glare and slightly conspicuous lens divisions keep it below a flawless 10.
A ultra high-resolution portrait of an elderly woman with silver hair and bifocal glasses, each lens reflecting a distant landscape, captured in warm late-afternoon sunlight - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 7 / 10 Rank 4 of 13: strong coverage of the requested elements gives this an advantage over several more natural portraits. The conspicuously banded reflection treatment is a substantial realism limitation.
An elephant made of clouds floating in a sunset sky, rendered in a surrealist photorealistic style with ethereal golden lighting and atmospheric depth. - GPT Image 2
GPT Image 2 8 / 10 Ranks third for its unusually convincing integration of elephant and natural cloudscape. Reduced facial separation and a less distinct overall silhouette prevent it from matching the two leaders.
An elephant made of clouds floating in a sunset sky, rendered in a surrealist photorealistic style with ethereal golden lighting and atmospheric depth. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 5 / 10 Ranks eleventh of 13 despite attractive lighting and detailed rendering. Its predominantly wool-like material, including the fibrous tusks, misses the cloud-texture requirement too substantially for a higher score.
A technology magazine cover featuring the headline "Tech Innovations of 2025," clean modern design style. - GPT Image 2
GPT Image 2 9 / 10 Ranked 3rd of 13, this is a highly convincing professional magazine cover. It trails the top two primarily because its denser product collage is less restrained and spatially clear.
A technology magazine cover featuring the headline "Tech Innovations of 2025," clean modern design style. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 6 / 10 The rendering is polished and the title is accurate, but the magazine-cover interpretation remains underdeveloped. Its poster-like presentation and generic visual treatment outweigh its otherwise strong technical execution.
Photorealistic aerial photograph clearly showing the edge of the Earth from space, capturing realistic curvature, atmosphere, and sunlight reflections - GPT Image 2
GPT Image 2 9 / 10 This is among the strongest images for coherent lighting and believable orbital photography. Softer foreground detail and a more conventional layout place it below the two leading examples.
Photorealistic aerial photograph clearly showing the edge of the Earth from space, capturing realistic curvature, atmosphere, and sunlight reflections - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 6 / 10 Ranks 9th of 13, with competent detail and an attractive large-scale composition. The weak sunlight reflection, exaggerated atmospheric treatment, and rendered-looking clouds prevent a higher realism-based score.

More comparisons