GPT Image 1 OpenAI Qwen Image Edit 2509 Alibaba

Settled by community votes across 7 shared challenges, with an AI judge weighing in on each.

GPT Image 1

22.6 arena score

#32 of 62 in Text-to-Image

Skill signature · Text-to-Image

Qwen Image Edit 2509

7.8 arena score

#32 of 32 in Image Editing

Vote tally

Where the votes landed

GPT Image 1

win rate

Ties

Qwen Image Edit 2509

win rate

Shared challenges 7

Challenge by challenge

The strongest take from each model on every shared challenge, with the AI judge's read.

Man and Car in California

Editing

Edit instruction

“Make a photo of the man driving the car down the California coastline”

Source

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Excellent preservation of the car's specific design details
+ Successfully combines both the man and the car from the source images
+ Higher realism and better lighting integration

− The man's hairstyle choice from the source is slightly altered in structure
− Motion blur on the wheels is a bit inconsistent with the car's body

Qwen Image Edit 2509

+ Beautiful cinematic landscape composition of the California coastline
+ Good sense of speed with realistic wheel motion and road blur

− The man is very small and lacks detail compared to the source image
− The car's proportions feel slightly elongated compared to the source

Verdict: GPT Image 1 is the clear winner as it maintains a much higher level of fidelity to both source images, making the man clearly identifiable and keeping the car's geometry accurate. Qwen Image Edit 2509 produces a visually stunning landscape, but the subject (the man) is barely visible and loses most of his specific features from the source.

Pose & Character Mashup

Editing

Edit instruction

“Use Image 1 as the exact pose reference and Image 2 as the character reference. Recreate the person/character from Image 2 in the exact dynamic pose and body position from Image 1. Keep the exact face, hair, clothing style/details, and expression from Image 2. Match the lighting and environment of Image 1. The final image must show the character from Image 2 performing the precise action/pose from Image 1 with perfect anatomy and natural integration.”

Source

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Followed the character reference by including the specific face, sunglasses, and black clothing with the white scarf scarf.
+ Accurately reproduced the complex pose and environment from the reference image.
+ Successfully integrated the character's features into the dynamic lighting and yellow background.

− The facial anatomy and expression look slightly distorted compared to the reference.
− The scarf physics and placement look somewhat stiff.

Qwen Image Edit 2509

+ Maintained very high visual quality and smoothness in the render.

− Completely failed the instructions to use the character from Image 2.
− Simply provided a slightly filtered version of Image 1 with no character swap.
− Ignored the core task of the prompt.

Verdict: GPT Image 1 followed all instructions, successfully swapping the character from Image 2 into the pose of Image 1 while maintaining his facial features and clothing details. Qwen Image Edit 2509 failed the challenge entirely by returning the original pose reference image with no discernible edits toward the character reference.

Outfit Transfer Challenge

Editing

Edit instruction

“Use Image 1 as the base person. Dress them in the exact elaborate outfit from Image 2 (including all layers, accessories, jewelry, and shoes). Carefully adapt the clothing to the body shape and pose in Image 1 while maintaining realistic fabric behavior, correct proportions, and perfect lighting/shadow matching. Keep the person’s exact face, hair, and background completely unchanged.”

Source

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Excellent adherence to the 'source preservation' constraint by keeping the face and features identical to Image 1.
+ High visual quality with realistic fabric textures and lighting that matches the beach environment.
+ Great attention to detail on the scarf pattern and coat texture.

− Cropped the original vertical aspect ratio into a square, losing the bottom half of the requested outfit (jeans and shoes).
− Slightly smoothed out some of the sand texture on the subject's face.

Qwen Image Edit 2509

+ Successfully included the full outfit including jeans and shoes as requested.
+ Maintains the pose and background of Image 1 reasonably well.

− Failed to preserve the face of the person in Image 1, instead creating a hybrid face that looks more like the person in Image 2.
− Lower visual quality with noticeable AI artifacts, particularly around the hands and where the subject meet the sand.
− The lighting on the subject is overly harsh and does not naturally match the sky and background.

Verdict: GPT Image 1 followed the most critical instruction of keeping the subject's face identical, resulting in a much more believable edit, even though it cropped the image and missed the shoes. Qwen Image Edit 2509 included the full outfit but failed the primary constraint of preserving the original person's face and produced an image with significantly more artifacts and poor lighting integration.

Bald man challenge

Image Editing

Edit instruction

“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Excellent preservation of original lighting and facial details
+ The hair color matches the existing beard perfectly
+ Maintains the rugged aesthetic of the source image

− The hairline is a bit high and stiff
− The hair shape is slightly blocky

Qwen Image Edit 2509

+ Natural and dynamic hair flow with realistic waves
+ The hairline integration is very smooth and believable
+ Superior rendering of individual hair strands and texture

− Slightly altered the tone of the skin on the forehead area

Verdict: Qwen Image Edit 2509 is the winner as it produces a much more natural-looking head of hair with realistic volume and movement. While GPT Image 1 successfully preserved most of the source image, the hair it added looks more like a wig compared to the organic integration seen in the Qwen output.

Over-the-top cartoon caricature

Editing

Edit instruction

“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.”

Source

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Excellent hand-drawn watercolor caricature aesthetic.
+ Highly accurate preservation of the subject's features and clothing from the source image.
+ Seamlessly integrates all three requested elements: news anchor desk, pet dog, and hockey.

− The composition is a bit more static than Model B.

Qwen Image Edit 2509

+ Cleverly integrates the news desk with a hockey rink setting.
+ Maintains the 'selfie' arm pose from the original source image.
+ Good use of vibrant, comic-book style colors.

− Gibberish text in the speech bubble and on screen.
− The facial likeness is less accurate to the source than Model A.
− Added a random second person that wasn't requested.

Verdict: GPT Image 1 (Model A) provides a much more professional caricature that remains true to the original subject's facial structure while perfectly balancing the requested themes. Qwen Image Edit 2509 (Model B) has an interesting creative concept of a hockey-rink studio, but is held back by nonsensical text and a lower-quality art style.

Studio Ghibli Anime Style

Editing

Edit instruction

“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”

Source

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Captures the Ghibli hand-painted aesthetic perfectly with watercolor textures and soft edges.
+ Excellent use of warm, nostalgic lighting and a dreamy, atmospheric background.
+ Preserves the iconic expressions and composition of the meme while translating them into the requested art style.

− The detail on the shirt is significantly simplified compared to the source image.
− The foreground woman's eyes are mostly closed, which is a slight departure from the original's profile/blurred gaze.

Qwen Image Edit 2509

+ Retains high detail from the source, such as the specific plaid pattern on the shirt.
+ Very high edit accuracy regarding character positioning and the background environment.
+ Clean linework that captures a more modern anime style.

− The style feels more like a standard digital manhwa or light anime rather than the specific 'hand-painted Ghibli' aesthetic requested.
− The colors are a bit too vibrant and lack the soft, pastel, nostalgic mood mentioned in the prompt.

Verdict: GPT Image 1 is the clear winner as it successfully interprets the 'Studio Ghibli' request with soft watercolor textures, warm lighting, and a distinct hand-drawn feel. While Qwen Image Edit 2509 is more accurate in preserving exact details like the shirt pattern, its style is too clean and digital, failing to capture the dreamy and nostalgic atmosphere essential to the Ghibli brand.

Golden Hour Stroll

Image Editing

Edit instruction

“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”

GPT Image 1

Qwen Image Edit 2509

AI Judge Analysis

GPT Image 1

+ Excellent source preservation, maintaining the woman's face and the dog's features perfectly
+ Realistic hair movement that blends naturally with the lighting
+ Subtle motion blur on the falling leaves enhances the sense of movement

Qwen Image Edit 2509

+ Adds colorful, vibrant leaves that create a strong visual contrast
+ Clear and distinct hair blowing effect

− Slightly alters the woman's facial features, making her look different from the source image
− The hair strands appear somewhat painterly and less integrated with the original texture
− Leaves look like 2D overlays rather than physical objects in the scene

Verdict: GPT Image 1 is the superior edit because it manages to add the requested motion effects while perfectly preserving the identity of the woman and the dog from the source image. Qwen Image Edit 2509 adds more vibrant elements, but there is a noticeable shift in the facial features and the hair texture looks less realistic compared to the original.