OpenAI
1094 Elo ±45
Rank #13 of 42 · median 37s per image · 11 refused
23–11–55
wins – ties – losses for GPT Image 1.5
89 direct comparisons
Elo expectation: GPT Image 1.5 wins 31%
Alibaba
1236 Elo ±57
Rank #9 of 42 · median 77s per image

By category

CategoryGPT Image 1.5Qwen-Image-3.0-ProDirect record
Anime & Cartoon Style 1128 1207 2–2–4
Architecture & Interiors 1160 1415 2–0–8
Complex Scenes 1070 1326 1–1–7
Ghibli style 1204 1270 1–1–1
Graphic Design 930 1180 2–1–7
Hands & Anatomy 1114 1213 3–1–5
Photorealistic People & Portraits 1108 1127 4–2–4
Surreal & Creative Prompts 1166 1234 3–1–6
Text in Images 1003 1318 0–2–8
Ultra Hard 1286 1310 5–0–5

By criterion

CriterionGPT Image 1.5Qwen-Image-3.0-Pro
Prompt adherence 1135 1157
Technical quality 1034 1228
Artistic merit 1079 1197
Detail execution 1072 1230
Realism & coherence 1006 1185

Side by side

A cartoon of a cat and a dog on an adventure, drawn in a colorful 2D style. - GPT Image 1.5
GPT Image 1.5 8 / 10 This is one of the stronger interpretations for explicit storytelling, vibrant styling, and plentiful well-executed details. Slight visual overcrowding keeps it below the most compositionally controlled images.
A cartoon of a cat and a dog on an adventure, drawn in a colorful 2D style. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 10 / 10 The strongest image in the set combines inventive worldbuilding, clear adventure storytelling, and fully resolved 2D craftsmanship. Its composition, character construction, prop handling, and environmental detailing reach an exceptional professional illustration standard.
Cross-section rendering of a multi-level underground bunker with different functional areas, structural support systems, ventilation, and power generation facilities visible. - GPT Image 1.5
GPT Image 1.5 6 / 10 An attractive and broadly relevant illustration, but its engineering organization is weaker than its visual presentation. Ambiguous circulation and loosely resolved machinery materially reduce its value as a technical cross-section.
Cross-section rendering of a multi-level underground bunker with different functional areas, structural support systems, ventilation, and power generation facilities visible. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 This is the most visually compelling rendering and a close competitor to the strongest architectural diagram. Its slightly more speculative machinery keeps it from taking the technical lead.
A bustling market scene with dozens of people buying and selling goods. - GPT Image 1.5
GPT Image 1.5 6 / 10 Vivid and highly faithful to the prompt, but weaker on the close-up interaction that should anchor the scene. The ambiguous weighing apparatus and soft hand structure outweigh its attractive atmosphere.
A bustling market scene with dozens of people buying and selling goods. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 One of the two strongest entries, this image excels at foreground interactions and finished presentation. Its promotional, slightly posed atmosphere makes it marginally less convincing as spontaneous bustle than the leading Thai market.
A kitchen with magical cooking implements preparing breakfast. Duplicate the artistic style of Studio Ghibli's domestic scenes, with Miyazaki's detailed settings and warm lighting. - GPT Image 1.5
GPT Image 1.5 7 / 10 Ranked 6th of 11, this delivers abundant magical cooking and an appealing breakfast atmosphere. The misdirected teapot pour is a major coherence flaw that prevents an excellent rating.
A kitchen with magical cooking implements preparing breakfast. Duplicate the artistic style of Studio Ghibli's domestic scenes, with Miyazaki's detailed settings and warm lighting. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Ranked 1st of 11, this most successfully integrates autonomous breakfast preparation, environmental richness, and warm lighting. Minor small-object softness and the somewhat realistic finish keep it below a flawless 10.
Minimalist logo for a sustainable coffee brand 'Evergreen Brew', single leaf motif, clean lines, vector style, white background. 'Evergreen Brew' contained as text in logo - GPT Image 1.5
GPT Image 1.5 3 / 10 Ranked 13th of 13, this is the weakest interpretation because both presentation and readability suffer. The recognizable leaf and correct text cannot compensate for the wrong background and unsuitable effect-heavy styling.
Minimalist logo for a sustainable coffee brand 'Evergreen Brew', single leaf motif, clean lines, vector style, white background. 'Evergreen Brew' contained as text in logo - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Ranked 1st of 13, this is the strongest balance of prompt fidelity, clarity, and practical logo design. Its familiar concept stops short of exceptional originality, but its execution is cleaner and more economical than the alternatives.
A group of five people joining hands in a circle, seen from above. - GPT Image 1.5
GPT Image 1.5 4 / 10 Ranked twelfth because it has only four people and a major hand-ownership problem at the center. Attractive color and sharp fabrics cannot rescue the prompt's primary anatomical interaction.
A group of five people joining hands in a circle, seen from above. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Ranked second, with the strongest atmosphere and environmental detail in the collection. Its more complicated finger layering places it just below image 3807 on the prompt's central anatomical challenge.
A 4K photo of an old fisherman with a weathered face and a pipe, grinning. - GPT Image 1.5
GPT Image 1.5 6 / 10 Ranks seventh for memorable expression and strong subject coverage. The visibly exaggerated skin treatment is a substantial limitation for a prompt specifically testing photographic age realism.
A 4K photo of an old fisherman with a weathered face and a pipe, grinning. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Ranks first for its combination of natural expression, convincing anatomy, and polished environmental storytelling. It has fewer conspicuous processing artifacts than the more aggressively textured portraits.
A waterfall that pours out stars and galaxies instead of water, illustrated in a cosmic fantasy style with vibrant space colors and dramatic lighting effects. - GPT Image 1.5
GPT Image 1.5 6 / 10 Rank 9 of 13. The prompt is fulfilled directly and the impact is dramatic. Excessive particle density, flattened depth, and an overbright base materially reduce the quality of execution.
A waterfall that pours out stars and galaxies instead of water, illustrated in a cosmic fantasy style with vibrant space colors and dramatic lighting effects. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Rank 1 of 13. This image achieves the strongest combination of literal prompt fulfillment, controlled spectacle, and environmental integration. Slight motif repetition and deeply compressed corner shadows keep it below a flawless score.
A storefront sign that says "Open 24/7" in neon lights, nighttime urban photography. - GPT Image 1.5
GPT Image 1.5 6 / 10 Rank 9 of 13: the correct text and attractive color scheme are clear strengths. Exaggerated weathering and less convincing luminous materials substantially weaken the photographic requirement.
A storefront sign that says "Open 24/7" in neon lights, nighttime urban photography. - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 Rank 1 of 13: this is the strongest combination of exact lettering, photographic credibility, and environmental storytelling. Minor simplification inside the diner keeps it below a flawless 10.
Photorealistic image of a robot painting a realistic self-portrait (i.e. the robot) on canvas, mimicking Van Gogh’s art style; clear, realistic metallic textures and painting details visible - GPT Image 1.5
GPT Image 1.5 5 / 10 More imaginative and self-portrait-aware than the human-portrait variants, but it fails the photorealistic outer-scene requirement. Its high painterly merit cannot compensate for that central stylistic mismatch.
Photorealistic image of a robot painting a realistic self-portrait (i.e. the robot) on canvas, mimicking Van Gogh’s art style; clear, realistic metallic textures and painting details visible - Qwen-Image-3.0-Pro
Qwen-Image-3.0-Pro 9 / 10 The strongest overall result, combining faithful prompt interpretation with particularly believable paint application and readable metallic surfaces. Minor synthetic neatness prevents a flawless score.

More comparisons