Black Forest Labs
1288 Elo ±51
Rank #6 of 42 · median 36s per image · 10 refused
Not judged in the same call yet
Elo expectation: FLUX 3 Image wins 75%
OpenAI
1094 Elo ±45
Rank #13 of 42 · median 37s per image · 11 refused

By category

CategoryFLUX 3 ImageGPT Image 1.5Direct record
Anime & Cartoon Style 1189 1128 —
Architecture & Interiors 1360 1160 —
Complex Scenes 1306 1070 —
Ghibli style 1472 1204 —
Graphic Design 1229 930 —
Hands & Anatomy 1315 1114 —
Photorealistic People & Portraits 1319 1108 —
Surreal & Creative Prompts 1371 1166 —
Text in Images 1272 1003 —
Ultra Hard 1278 1286 —

By criterion

CriterionFLUX 3 ImageGPT Image 1.5
Prompt adherence 1230 1135
Technical quality 1259 1034
Artistic merit 1175 1079
Detail execution 1271 1072
Realism & coherence 1234 1006

Side by side

A magical girl character in a frilly dress, casting a spell (classic anime trope). - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranked first for its combination of unmistakable trope adherence, clean anatomy, and exceptionally readable spellcasting choreography. It is the most balanced entry, with only minor limitations in background variety and originality.
A magical girl character in a frilly dress, casting a spell (classic anime trope). - GPT Image 1.5
GPT Image 1.5 7 / 10 Ranked 5th of 13, this offers excellent prompt fulfillment and considerable visual appeal. Overcrowding and uneven resolution of the glove and ruffles make it less polished than its initial sparkle suggests.
A Gothic cathedral interior at sunset with light streaming through detailed stained glass windows, creating colored patterns on stone floors and columns. Show accurate ribbed vaulting. - FLUX 3 Image
FLUX 3 Image 9 / 10 This ranks second through its exceptionally coherent architecture and comprehensive treatment of the requested lighting effects. Its more conventional composition and simpler floor patterns place it just behind the strongest image.
A Gothic cathedral interior at sunset with light streaming through detailed stained glass windows, creating colored patterns on stone floors and columns. Show accurate ribbed vaulting. - GPT Image 1.5
GPT Image 1.5 6 / 10 Ranks ninth for dramatic atmosphere and generally competent Gothic architecture. Overprocessed lighting and less convincing material response prevent the visual spectacle from becoming a high-coherence result.
A bustling market scene with dozens of people buying and selling goods. - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranked 1st of 13, this image offers the strongest combination of crowd complexity, readable commerce, and controlled composition. Minor simplification in small anatomical details prevents a flawless score.
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 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. - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranks first of eleven: it provides the strongest combination of explicit magical preparation, detailed domestic world-building, warm lighting, and technical control. Minor overlap ambiguities prevent a flawless score, but it is the most complete response to the prompt.
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.
App icon design for a weather app. Minimalist representation of a sun partially behind a stylized cloud, subtle gradient, modern flat style. - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranked 1st of 13, this is the strongest combination of prompt fidelity, clean execution, and small-format clarity. Its familiar visual vocabulary keeps it just short of an exceptional, uniquely memorable result.
App icon design for a weather app. Minimalist representation of a sun partially behind a stylized cloud, subtle gradient, modern flat style. - GPT Image 1.5
GPT Image 1.5 4 / 10 Ranked 13th of 13. It preserves the weather subject, but its dominant lighting effects and layered rendering are the largest departures from the requested application and style.
A group of five people joining hands in a circle, seen from above. - FLUX 3 Image
FLUX 3 Image 9 / 10 This is the strongest image in the set because it combines exact prompt adherence with convincing hand interactions. Its clear five-person ring distinguishes it from the numerous incorrect counts and central hand piles.
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 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 - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranked 1 of 13, this offers the best balance of prompt specificity, photographic credibility, and expressive portraiture. Its clearly indicated bifocals and restrained landscape reflections distinguish it from the more conspicuously composited alternatives.
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 1.5
GPT Image 1.5 6 / 10 Rank 7 of 13: this is visually arresting and richly detailed, but weaker in physical optics. The postcard-like reflections and absent visible bifocal segments substantially limit its realism and prompt accuracy.
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. - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranked second of 13, this is one of the most direct and visually spectacular interpretations of the prompt. A slightly ambiguous lower basin places it just behind the more completely integrated leading image.
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 motivational poster with the quote "Dream Big, Work Hard" in stylized font, minimalist graphic design. - FLUX 3 Image
FLUX 3 Image 9 / 10 Ranked 1st of 13 for the strongest combination of expressive typography, readability, and restrained decoration. It looks professionally usable, with only slightly tight spacing between the two text lines.
A motivational poster with the quote "Dream Big, Work Hard" in stylized font, minimalist graphic design. - GPT Image 1.5
GPT Image 1.5 6 / 10 Ranks ninth despite being one of the strongest standalone illustrations in the set. Its central mismatch with the minimalist requirement substantially outweighs its impressive craftsmanship for this task.
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 - FLUX 3 Image
FLUX 3 Image 7 / 10 Rank 5 of 13. Its excellent photographic execution makes it the strongest partially compliant interpretation, but the human-cyborg portrait significantly weakens the essential self-likeness.
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.

More comparisons