GPT Image 1.5 OpenAI vs Qwen Image 2.0 Alibaba

GPT Image 1.5

• + Perfectly captures the physics of the scene with the sphere resting on the bottom surface.
• + Refractions through the glass cube and reflections on the table are highly realistic.
• + The lighting accurately comes from the left as requested, creating soft shadows.

• − The plant in the background is quite dense, which slightly obscures the clarity of the glass cube concept.

Qwen Image 2.0

• + Excellent texture work on the book cover and the pages of the book.
• + The plant is well-positioned and clearly visible through the glass panels.

• − The sphere is floating in the center of the cube without any support, which feels unnatural for the requested scene.
• − The reflection/refraction of the sphere appears duplicated and physically inconsistent on the side panels.

GPT Image 1.5

• + Excellent photographic atmosphere with wet textures and rain visibility.
• + Stronger cinematic lighting and reflection quality on the pavement.
• + Anatomy and tool interactions are highly realistic.

• − The car in the background is quite sharp, lacking the requested motion blur.
• − The framing feels a bit too balanced for 'imperfect framing'.

Qwen Image 2.0

• + Better adherence to 'imperfect framing' with a more candid, snap-shot composition.
• + Skin textures are highly detailed and naturalistic.
• + Includes motion blur on the passing vehicle as requested.

• − Physical logic errors with the bicycle, such as the chain and pedal placement.
• − The man's hands have structural inconsistencies (too many fingers/blended joints).

GPT Image 1.5

• + Exceptional photographic realism in skin texture and eyes
• + Masterful use of warm lighting and bokeh to create atmosphere
• + Incredibly detailed engraving on the armor and texture on the fabric/leather

• − The hair beads are somewhat simple and look slightly fused with the hair in places

Qwen Image 2.0

• + Strong character design with distinct beaded braids
• + Good representation of ornate armor and battle scars
• + Vibrant colors and clear background story elements

• − The lighting looks artificial and does not realistically reflect off the skin
• − Anatomical issues with the hand holding the sword pommel
• − Eyes have an unnatural, glowing quality that detracts from realism

GPT Image 1.5

• + Excellent text readability and logical content
• + High-quality, appetizing food photography
• + Professional UI/UX layout that follows standard menu design principles

• − The grid of images is slightly asymmetrical compared to the text blocks

Qwen Image 2.0

• + Strong minimalist aesthetic with a consistent image grid
• + Appearing more like a modern digital app menu

• − Gibberish text making the menu unusable
• − Incorrect categorization, such as showing pizza under the 'Mains' column
• − Repetitive pricing and lack of descriptive text

GPT Image 1.5

• + Excellent photorealistic texture on the meat and bun.
• + Vibrant and dynamic composition with sparks and embers filling the space.
• + Clear and professional typography that integrates well with the theme.

• − The 'exploded' effect is slightly more clumped than floating individual layers.

Qwen Image 2.0

• + Strong 'exploded' burger effect with clear separation between components.
• + Effective use of literal fire within the text design.
• + Good adherence to the starburst pricing requirement.

• − The burger ingredients look slightly more synthetic and less appetizing than Model A.
• − The background feels more like a flat black backdrop with some smoke rather than a fiery environment.

GPT Image 1.5

• + Perfect text accuracy for all requested items and prices.
• + Highly realistic chalk texture with dusty smudges and varying pressure.
• + Excellent handwriting consistency that feels human rather than a font.

• − Lacks the environmental context of the 'cozy café' mentioned in the prompt, focusing only on the board.
• − Slightly cramped composition toward the bottom.

Qwen Image 2.0

• + Successfully includes the 'cozy café' background with depth of field.
• + Strong composition making use of the board's physical presence in a space.
• + Accurate spelling and character rendering.

• − The text layout is slightly awkward with large gaps before the prices.
• − The handwriting looks a bit more like a digital brush than physical chalk compared to the other model.

GPT Image 1.5

• + Excellent cinematic lighting and texture on the spacesuit and horse fur.
• + High level of detail in the lunar surface and background celestial bodies.
• + Dynamic composition with a sense of motion and grit.

• − The astronaut is on top of the horse, failing the specific spatial instruction 'horse on top, not vice versa'.

Qwen Image 2.0

• + Beautiful surreal atmosphere with shimmering horse skin and floating droplets.
• + Clean, sharp rendering of the astronaut and planetary background.
• + Better usage of negative space to create a 'space' feel.

• − Fails the specific prompt instruction 'horse on top, not vice versa'.
• − Anatomical issues with the horse's front legs.

GPT Image 1.5

• + Excellent photorealistic lighting and textures in the capybara's fur and jacket.
• + Stronger cinematic composition that captures both characters clearly.
• + Superior interior detailing, including the dashboard meter and realistic upholstery.

• − The passenger's eyes and facial features are slightly distorted under zoom.

Qwen Image 2.0

• + Clearer view of the capybara's paws on the steering wheel.
• + Good background bokeh effect with Manhattan city lights.

• − The passenger is seated too far forward, appearing to be in the front seat or mid-air rather than the back seat.
• − The blue car seats and interior feel less like a traditional New York taxi cabinet.
• − Noticeable anatomy artifact on the passenger's left hand.

GPT Image 1.5

• + Excellent typography style that feels integrated into the vintage aesthetic.
• + Beautiful cinematic lighting and atmospheric painterly quality.
• + Perfect adherence to all requested text and visual elements.

• − The 'Date' text at the bottom is slightly smaller and more compressed than other text elements.

Qwen Image 2.0

• + Clear, legible text rendering across all sections.
• + Strong central composition with a bold jack-o-lantern.
• + Includes nice stylistic gothic fonts for the heading.

• − The background and foreground elements feel somewhat disjointed, like a digital collage.
• − The border appears slightly more generic compared to the intricate thorns in Model A.
• − Lighting is flatter and less 'cinematic' than the other version.

GPT Image 1.5

• + Expertly captures the isometric miniature diorama aesthetic with a layered base.
• + Excellent text rendering with clean, stylized typography centered as requested.
• + Rich, high-quality PBR-like textures on the wooden board and ceramic teapot.

• − Includes more elements (teapot, soy sauce) than the 'minimal' request asked for.
• − The flag icon is slightly simple compared to the high-detail 3D scene.

Qwen Image 2.0

• + Follows the 'minimal garnish and plate' instruction closely.
• + Accurate 45-degree angle and solid blue background.
• + Clear, bold text centered at the top.

• − Lacks the 'miniature 3D cartoon' style, looking more like a standard product photo.
• − The 'diorama base' is just a simple cutting board rather than a stylized scene.
• − The text layout is less integrated and looks like a basic overlay.

GPT Image 1.5

• + Features very high-quality lighting with distinct god rays and warm golden tones.
• + Captures the 'big expressive eyes' and 'tumbling' aspects of the prompt very effectively.
• + Includes all requested animals in a harmonious, intimate composition.

• − The fox kit has the same facial lighting and structure as the puppy, making them look slightly too similar.
• − The butterfly appears slightly flat against the background compared to the 3D quality of the animals.

Qwen Image 2.0

• + Excellent action-oriented composition that looks more like a natural play session.
• + Realistic fur textures and distinct anatomical features for each of the four animals.
• + Better depth of field and variety in the wildflower meadow.

• − The fox kit is poorly positioned and appears to have an anatomical issue where its head meets its body.
• − The background lighting is a bit hazy and lacks the crisp 'masterpiece' feel of the animals in the foreground.

GPT Image 1.5

• + Excellent typography with a sophisticated, professional font style.
• + Accurate use of a banner and a secondary banner-like accent.
• + Good vector-style shading and texture application.

• − Ignored the request for a light background, providing a black one instead.
• − The steam element is somewhat detached and small.

Qwen Image 2.0

• + Correctly used a light background as specifically requested.
• + The cloche is central and well-integrated into the design.
• + Text is clear and highly legible.

• − The steam looks more like fire and is placed awkwardly inside the cloche handle area.
• − The banner on the bottom has an illogical fold/curl on the right side.
• − Typography is very basic and lacks the 'vintage' feel of a high-end cafe.

GPT Image 1.5

• + Excellent layout with distinct, sequential panels for each step.
• + Highly legible text with perfect spelling for all required stages.
• + Consistent and professional vector illustration style with a clear color palette.

• − Slightly repetitive background elements in the space panels.
• − Translunar panel shows both a tiny moon and a large moon which is a bit confusing.

Qwen Image 2.0

• + Elegant vertical layout that mimics a traditional poster.
• + Includes additional thematic elements like the names of the astronauts in the corner.
• + Clean, minimalistic design that adheres well to a modern NASA-inspired palette.

• − Spelling error in 'Translunjar'.
• − The Descent and Landing labels overlap the corresponding illustrations awkwardly.
• − Visual hierarchy is slightly weaker with elements floating on a single background.