FLUX.2 [max] vs GPT Image 1.5
Head-to-head across 16 challenges
FLUX.2 [max]
20.0%
win rate
Ties
20.0%
GPT Image 1.5
60.0%
win rate
Challenge Results
Geometric Composition
Text-to-Image“A glass cube on a wooden table. Inside the cube is a small blue sphere. On top of the cube sits a red book. A green plant is behind the cube, partially visible through the glass. Soft window light from the left.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent cinematic lighting with realistic bokeh and shadows.
- + Very high texture detail, particularly on the table grain and the leather-like book cover.
- + Correct interpretation of the 'partially visible through glass' plant element.
- − The glass cube lacks a physical top panel, making the book appear to float slightly or rest only on the edges.
- − The plant in the background is significantly blurred compared to the main subject.
GPT Image 1.5
- + Very clean and solid construction of the glass cube with clear edges and a visible top surface.
- + Strong adherence to all prompt elements including the lighting direction and object placement.
- + Realistic reflections on the blue sphere and the bottom of the cube.
- − The wood grain on the table is somewhat generic compared to Model A.
- − The plant lacks the depth and soft focus that would make the composition feel more professional.
Verdict: Both models followed the prompt perfectly, including the complex spatial relationships between the objects and the specific lighting direction. FLUX.2 [max] produced a more artistic, high-end photographic result with superior textures, but GPT Image 1.5 handled the physical logic of the glass cube better by ensuring the book had a clear surface to rest upon.
Man and Car in California
Editing“Make a photo of the man driving the car down the California coastline”
AI Judge Analysis
FLUX.2 [max]
- + Excellent preservation of the car's model and specific design details.
- + Accurately places the man inside the car while maintaining his scarf and dreadlocks.
- + High-quality environment and lighting integration.
- − The man's expression is quite stern compared to the original smile.
- − The scaling of the man seems slightly small relative to the car's interior.
GPT Image 1.5
- + Captures the man's facial expression and smile more accurately than Model A.
- + Great dynamic composition showing the perspective of driving along the coast.
- − Significant loss of car identity, changing the grill and front details of the Rolls Royce.
- − Inaccurate hand placement/anatomy on the steering wheel.
Verdict: FLUX.2 [max] is the superior model for this edit because it preserves the specific identity of both source subjects: the exact Rolls Royce model and the man's distinct style. While GPT Image 1.5 captures the man's smile better, it fails to maintain the car's design and features mangled hand geometry.
Candid Street Photography
Text-to-Image“A candid street photo of an elderly Japanese man repairing a red bicycle in light rain, reflections on wet pavement, shallow depth of field, 50mm lens, natural skin texture, imperfect framing, motion blur from passing cars, cinematic but realistic, no stylization.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent adherence to the motion blur request for passing cars.
- + Highly realistic skin textures and fine details on the hands.
- + Accurately captures the 'imperfect framing' of a candid street photo.
- − The brake cables on the bicycle are a bit messy and structurally illogical.
- − The background cars have slightly surreal lighting/ghosting.
GPT Image 1.5
- + Stronger atmospheric rain effects with visible droplets on the clothing and hat.
- + Great attention to the reflections on the wet pavement.
- + Composition feels very grounded and lifelike.
- − Missed the specific request for motion blur on the passing cars; the background car is static.
- − Minor anatomical confusion where the hands meet the bicycle chain area.
Verdict: FLUX.2 [max] followed the complex technical prompt more closely, particularly by including the requested motion blur on background traffic and the imperfect framing. While GPT Image 1.5 delivered a more atmospheric rain effect and beautiful reflections, it failed to incorporate the motion blur, resulting in a more static scene. FLUX.2's skin textures and adherence to the 50mm lens look make it the more successful interpretation of the 'candid street photo' aesthetic.
Modern Clean Menu
Text-to-Image“Modern minimalist restaurant menu design, white background with colorful food photos in grid, sections for appetizers/pizza/mains, bold sans-serif fonts, vibrant accents, clean professional layout for casual dining.”
AI Judge Analysis
FLUX.2 [max]
- + Strictly followed the request for a grid of food photos.
- + Excellent use of vibrant secondary colors (yellow, red, green) for section headers.
- + Clean, professional aesthetic that feels like a real modern layout.
- − The text is largely gibberish/placeholder text.
- − Some image artifacts present, like a floating hand in the slider photo.
GPT Image 1.5
- + Perfect English text and logical menu pricing.
- + Clear association between menu categories and the corresponding food photos.
- + High visual quality of the food photography.
- − The food photos are in stacked blocks rather than a true grid layout as requested.
- − Design feels a bit more generic compared to the bold styling of the other model.
Verdict: While FLUX.2 [max] captures the 'grid' and 'bold vibrant accents' of the prompt much more effectively, GPT Image 1.5 produces a functional menu with perfect English text. However, as a design challenge, FLUX.2 [max] feels more closely aligned with the requested modern minimalist aesthetic and specific layout requirements.
Bald man challenge
Editing“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent preservation of the original image's lighting, background, and clothing.
- + The hair texture is high-resolution and matches the existing beard well.
- + Maintains the exact facial geometry and expression of the source image.
- − The hair volume is arguably a bit exaggerated at the top, looking slightly like a wig.
- − The hair encroaches a bit too much on the ear/glasses contact point.
GPT Image 1.5
- + Natural, realistic hair density and curl pattern that fits the character's aesthetic.
- + Very clean integration of the hairline with the forehead.
- + Almost perfect preservation of pixels outside the hair area.
- − Slightly altered the shape of the glasses frames.
- − Subtle changes to the nose and eye area making the person look like a slightly younger version of themselves.
Verdict: Both models did an exceptional job at adding realistic hair while preserving the scene. FLUX.2 [max] is preferred because it maintained the person's exact facial features and the original glasses perfectly, whereas GPT Image 1.5 subtly face-swapped a slightly different person onto the head, despite having a more natural-looking hair style.
Night Sky Transformation
Editing“Change the scene to night: a deep, dark sky with subtle, glistening stars visible behind the mountain.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent preservation of the source image's geometry and fine details.
- + Realistic night lighting that convincingly darkens the sky while maintaining visibility of the village.
- + Subtle, natural-looking stars as requested.
- − The mountain peak still retains a slight warm 'sunset' glow on its face that contradicts the dark sky.
GPT Image 1.5
- + Conveys a very deep, dark night atmosphere.
- + Successfully replaces the sky with a dense field of stars.
- − Loss of detail in the mid-ground and village due to extreme underexposure.
- − Alters the shape of the mountain peak, making it look thinner and less like the iconic Matterhorn in the source.
- − The replacement of the village lighting is much more generic than the source image.
Verdict: FLUX.2 [max] is the winner because it successfully transforms the sky to night while perfectly preserving the intricate architecture of the village and the structural integrity of the mountain. GPT Image 1.5 creates a convincing dark atmosphere but fails as an editing model, as it significantly degrades the source image's resolution and alters the shape of the main subject.
Adorable Baby Animals in Sunny Meadow
Text-to-Image“Hyper-photorealistic scene of fluffy baby animals—a golden retriever puppy, tabby kitten, baby bunny, and red fox kit—with big expressive eyes and ultra-detailed soft fur, playfully chasing butterflies and tumbling together in a lush wildflower meadow, warm golden sunrise light with god rays and dew sparkles, joyful wholesome vibe, 8K masterpiece.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent composition with a sense of movement as the animals run through the field.
- + Very high level of realism in the fur texture and the dew drops on the grass.
- + Correct anatomical rendering of all four requested animals with distinct features.
- − The lighting is a bit hazy, which slightly reduces the 'pop' of the subjects.
- − The butterflies look a bit static compared to the dynamic movement of the animals.
GPT Image 1.5
- + Vibrant, warm lighting with very strong god rays and 'glittery' dew effects.
- + Captures the 'tumbling' and 'joyful' vibe perfectly with expressive facial expressions.
- + Extremely cute and stylized to emphasize the 'big expressive eyes' requested.
- − The puppy's left paw has five toes with prominent black pads that look slightly unnatural.
- − The animals are crowded together in a way that feels a bit less realistic than the spacing in Model A.
- − The fox's anatomy is slightly more generic/dog-like than the fox in Model A.
Verdict: FLUX.2 [max] provides a more sophisticated and realistic 8K masterpiece with better anatomical accuracy and a beautiful sense of depth. GPT Image 1.5 wins on pure emotional 'wholesomeness' and lighting effects, but it suffers from minor AI artifacts in the paws and a slightly more cluttered composition. FLUX.2 [max] is preferred for its cleaner technical execution and realistic integration of the four animals into the environment.
Over-the-top cartoon caricature
Editing“Create a caricature of me and my job. Make it exaggerated and humorous, incorporating my profession as a tv show anchor and my love for dogs and hockey.”
AI Judge Analysis
FLUX.2 [max]
- + Successfully incorporates all elements: TV anchor desk, dogs, and hockey background.
- + The character maintains a distinct resemblance to the person in the source image in a cartoon style.
- + Clean vector-style illustration with high clarity.
- − The 'caricature' style is more of a generic avatar/clipart style rather than an exaggerated caricature.
- − The dogs are repetitive in design.
GPT Image 1.5
- + Excellent caricature style with exaggerated features that still strongly resemble the source subject.
- + Highly creative integration of the dog wearing a hockey helmet and the 'Breaking News' ticker.
- + Rich, vibrant colors and dynamic composition that feels energetic and humorous.
- − Minor text rendering issues on the microphone and ticker, though largely legible.
- − The transition between the desk and the character's body is slightly cluttered.
Verdict: GPT Image 1.5 is the clear winner as it delivered a true humorous caricature with exaggerated features while maintaining a striking resemblance to the source image. Although FLUX.2 [max] adhered to all prompt instructions, its style felt more like a flat digital illustration than the requested 'exaggerated and humorous' caricature.
Victorian Greenhouse Oasis
Text-to-Image“Hyper-photorealistic interior of a lush Victorian glass greenhouse filled with exotic tropical plants, vibrant blooming orchids, tall ferns, colorful butterflies in flight, sunlight filtering through ornate glass roof creating realistic caustics and dew on leaves, intricate iron framework visible, misty atmosphere, 8K masterpiece.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent rendering of the intricate Victorian iron framework and glass tiling.
- + Well-balanced composition with a clear floor area and inviting park benches.
- + Subtle and realistic misty atmosphere that interacts with god rays.
- − Butterflies look somewhat pasted on and lack motion blur or realistic integration.
- − Orchids and plants feel slightly more generic and less 'lush' compared to the second image.
GPT Image 1.5
- + Excellent adherence to the 'dew on leaves' and 'misty atmosphere' prompts with visible water droplets.
- + Dynamic lighting with very strong volumetric god rays and realistic caustics.
- + Plants look more exotic and varied, creating a denser, more jungle-like interior.
- − The composition is a bit cluttered with no clear focal point in the middle ground.
- − Some butterfly wings have slightly unnatural geometry when viewed at full scale.
Verdict: Both models followed the prompt well, but GPT Image 1.5 captured the specific textural details like dew and mist with much more intensity, resulting in a more 'lush' feel. FLUX.2 produced a more structurally sound and clean architectural space, but it felt a bit drier and less atmospheric than requested. GPT Image 1.5 is the winner for its superior handling of light, moisture, and exotic plant variety.
Heroic Super Hero Portrait
Text-to-Image“Hyper-photorealistic full-body portrait of a female superhero standing triumphantly on a New York skyscraper rooftop at golden sunset, wearing a classic modest superhero costume with flowing cape, chest emblem, gloves, and boots in red and blue colors, practical design, short hair, strong determined heroic expression looking into the distance, powerful confident stance with hands on hips and cape billowing dramatically in the wind, detailed urban cityscape background, warm natural sunlight with sharp shadows and fabric highlights, ultra-sharp textures on suit, hair, and concrete, 8K masterpiece, empowering family-friendly style.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent full-body composition with a true sense of scale and perspective.
- + The costume design is more modern, detailed, and visually integrated.
- + Exceptional background detail, including realistic city traffic lighting and atmospheric perspective.
- − The cape movement is a bit stiff compared to the 'billowing' request.
GPT Image 1.5
- + The cape has a very dynamic and dramatic billowing effect as requested.
- + The 'short hair' requirement is interpreted more distinctly.
- + High clarity and vibrant color saturation.
- − Failed the 'modest' requirement by removing the pants found in traditional suits, opting for a short skirt/tunic.
- − The background lighting on the towers feels slightly flat and less integrated with the sunset.
Verdict: FLUX.2 [max] is the winner for its superior atmospheric integration and adherence to the 'modest' and 'practical' costume descriptors, creating a more professional cinematic look. While GPT Image 1.5 captured the billowing cape and hair length better, it deviated from the modesty request and the overall composition feels more like a studio composite than a coherent scene.
Neutral Expression to Genuine Smile
Editing{
"action": "image_edit",
"reference": "uploaded neutral portrait",
"change": "Warm genuine Duchenne smile: lips curved up, slight natural teeth, soft eye crinkles, subtle cheek raise",
"details": "Realistic smiling skin (dimples if present, soft cheek shadows), slightly brighter eyes; keep exact eye shape/color/iris",
"preserve_exact": "Face identity/structure, eyes/nose/lips/eyebrows, hair, skin texture/pores/freckles, makeup, clothing, head pose, background, lighting, shadows, framing",
"no_changes": "No face shape change, no new features, no gaze shift, no hair/clothing/lighting/background edits",
"style": "Ultra-photorealistic 8K portrait, sharp face focus, natural soft lighting, realistic skin glow"
}
AI Judge Analysis
FLUX.2 [max]
- + Successfully applied eye crinkles and a broad, genuine Duchenne smile
- + Highly realistic skin folding around the mouth and cheeks
- + Matches the requested 'subtle cheek raise' and brightness in the eyes
- − Slightly alters the shape of the chin/jawline compared to the source
- − The skin texture appears a bit waxy in the highlighted cheek areas
GPT Image 1.5
- + Excellent preservation of the original face shape and identity
- + Maintains nearly identical hair placement and background details
- + Clean, natural tooth rendering that matches the requested 'slight natural teeth'
- − The smile feels less 'genuine' or Duchenne because it lacks the requested eye crinkles
- − The expression is more of a polite smile than the warm, expressive change requested
Verdict: FLUX.2 [max] did a much better job of capturing the specific 'Duchenne' quality requested, including the eye crinkles and cheek shadows, though it slightly altered the facial structure. GPT Image 1.5 was better at preserving the original image's pixels and identity perfectly, but the smile modification was too subtle and failed to incorporate the eye crinkles.
Studio Ghibli Anime Style
Editing“Transform this photo into a Studio Ghibli–inspired illustration. Use soft pastel colors, hand-painted textures, gentle lighting, dreamy backgrounds, and a warm, nostalgic mood”
AI Judge Analysis
FLUX.2 [max]
- + Perfectly preserves the composition and poses of the original meme.
- + Art style is very close to authentic hand-drawn anime with subtle paper textures.
- + Maintains specific clothing details like the plaid pattern on the shirt accurately.
GPT Image 1.5
- + Excellent soft, dreamy lighting and pastel color palette.
- + Achieves a high level of aesthetic 'magic' associated with Ghibli backgrounds.
- + Captures the emotional expressions of the characters very well in an illustrative style.
- − The man's hand is oddly merged with his side/hip.
- − The girl in the red dress has much longer, more voluminous hair than the source image.
Verdict: Both models did an excellent job translating the 'distracted boyfriend' meme into a Ghibli-inspired style. FLUX.2 [max] is the winner for its incredible preservation of the source image's geometry and details, whereas GPT Image 1.5 took more creative liberties with the hair and lighting that slightly drifted from the original composition.
Intricate Floral Mandala
Text-to-Image“Perfectly symmetrical mandala made entirely of real flowers, petals, leaves, fruits, and seeds in vibrant natural colors, intricate layered patterns with radial symmetry, top-down view on a soft neutral background, hyper-detailed organic textures and subtle shadows, photorealistic, 8K masterpiece.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent photorealism with natural textures and realistic lighting.
- + Includes a good variety of items like sliced citrus and seeds as requested.
- + Features subtle, convincing shadows that give the objects depth.
- − Symmetry is slightly imperfect in the outer flower petals.
- − Less vibrant and colorful than the alternative.
GPT Image 1.5
- + Extremely high level of radial symmetry throughout the entire pattern.
- + Very vibrant and saturated colors that pop against the background.
- + Strong adherence to the layered mandala structure.
- − Lighting feels a bit flat and artificial compared to Model A.
- − Some elements, like the berries and leaves, have a slightly 'plastic' or computer-generated texture.
Verdict: FLUX.2 [max] wins on photorealism, providing organic textures and realistic lighting that make the composition feel like a real physical arrangement. While GPT Image 1.5 offers superior mathematical symmetry and more vibrant colors, it lacks the convincing '8K masterpiece' texture and depth found in the FLUX output.
Golden Hour Stroll
Editing“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
AI Judge Analysis
FLUX.2 [max]
- + Successfully added hair blowing in the wind while maintaining hair texture consistency.
- + Added floating leaves that appear somewhat natural to the scene.
- + Preserved the identity of the woman and the dog almost perfectly.
- − The 'flying' leaves are static and lack motion blur, feeling more like they are pasted on top.
- − A weird artifact appears on the woman's left hand where her fingers now look distorted.
GPT Image 1.5
- + Excellent addition of dynamic motion through motion-blurred leaves.
- + The hair blowing looks very natural and follows a logical wind direction.
- + The overall lighting and environment feel more 'energetic' as requested.
- − Slightly altered the woman's facial features compared to the source.
- − Several leaves overlap the dog and the woman's clothes in a way that looks like a digital overlay.
Verdict: Both models followed the instructions well, but GPT Image 1.5 captured the 'energetic and lively' feel much better by incorporating motion blur on the flying leaves. While FLUX.2 [max] preserved the source image details more accurately (aside from a hand glitch), its 'flying' leaves feel too static and disconnected from the environment.
Vintage Cafe Logo
Text-to-Image“Vintage minimalist restaurant logo for "Caffè Florian", retro cloche dome with steam and "Est. 1720" banner, classic typography, warm brown and cream tones, subtle texture on light background, vector emblem style.”
AI Judge Analysis
FLUX.2 [max]
- + Perfect text rendering for both name and date
- + Accurately follows the light background requirement with subtle texture
- + Excellent vector emblem composition that feels authentic for a vintage logo
- − The steam lines are a bit thin and could be more prominent
GPT Image 1.5
- + Strong artistic texture on the cloche dome
- + Dynamic typography style
- − Failed the light background prompt, using a pitch-black background instead
- − Rendering of the text 'FLORIAN' has minor inconsistencies in letter weight
- − The vector look is slightly more illustrative than a professional logo emblem
Verdict: FLUX.2 [max] followed the prompt much more accurately, especially regarding the light background and subtle texture requirements. While GPT Image 1.5 has an interesting artistic style, its failure to provide a light background and slight issues with font consistency make FLUX.2 [max] the clear winner for a professional logo design.
Apollo 11: Journey to Tranquility
Text-to-Image“Create a clean, modern vector infographic poster about the Apollo 11 mission. NASA-inspired palette (navy, white, muted red, light gray). Flat-vector style, crisp lines, consistent iconography, subtle gradients only. Steps (stop at landing): 1. Launch (Saturn Vicon) 2. Earth Orbit (Earth + orbit ring icon) 3. Translunar (trajectory arc icon) 4. Lunar Orbit (Moon + orbit ring icon) 5. Descent (lunar module descending icon) 6. Landing (lunar module on the surface icon) Small supporting elements (minimal text): • Crew strip: three silhouette icons with only last names: Armstrong, Aldrin, Collins. • Landing site marker: Moon pin labeled "Tranquility" only. Layout constraints: generous margins, large readable labels, clean background with subtle stars. Vector-only, print-poster look, high resolution.”
AI Judge Analysis
FLUX.2 [max]
- + Excellent typography with correct spellings for all names and labels.
- + Well-organized grid layout following the requested sequential steps.
- + Very clean flat-vector style with consistent iconography.
- − The sequence is slightly out of order, placing Lunar Orbit before Translunar.
- − Included yellow in the palette which wasn't specifically requested but fits the lunar module theme.
- − The 'Tranquiity' label has a minor typo.
GPT Image 1.5
- + Strong NASA-inspired color palette usage and dramatic vector styling.
- + Correct chronological ordering of all mission steps.
- + High-quality illustrations of the Saturn V and Lunar Module.
- − The image is cropped at the top, cutting off the title.
- − The 'Translunar' step features a rocket rather than a trajectory arc icon as requested.
- − The 'Earth Orbit' step shows the Earth from the same perspective as the launch step, feeling redundant.
Verdict: FLUX.2 [max] produced a much more professional and complete infographic layout with clear labels and a polished aesthetic, despite an ordering error in the steps. GPT Image 1.5 followed the chronological order better but ultimately failed as a poster due to the top being cut off and less consistent icon styles.
FLUX.2 [max]
Black Forest Labs' flagship image generation model delivering state-of-the-art quality with exceptional realism, precision, and consistency for both text-to-image and advanced image editing
GPT Image 1.5
OpenAI's state-of-the-art image generation model with better instruction following and adherence to prompts