ByteDance's image generation model with integrated text-to-image and image editing capabilities in a unified architecture, supporting up to 4K resolution
Settled by community votes across 19 shared challenges, with an AI judge weighing in on each.
Seedream 4.0
#15 of 62 in Text-to-Image
Wan 2.5 (Preview)
#27 of 62 in Text-to-Image
Where the votes landed
Seedream 4.0
0%
win rate
Ties
0%
Wan 2.5 (Preview)
0%
win rate
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
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
Seedream 4.0
- + Excellent adherence to the 'soft window light' lighting request.
- + Realistic glass reflections on the table and within the cube.
- + Clean, minimalist composition.
- − The glass enclosure is rectangular rather than a true cube.
- − The blue sphere is quite small relative to the space.
Wan 2.5 (Preview)
- + Perfect cube geometry for the glass container.
- + Highly detailed textures on the book cover and the plant leaves.
- + Better proportion between the sphere and the cube.
- − The sphere appears to be floating slightly above the bottom surface.
- − Unnecessary dust/particle effects in the air distract from the clarity of the scene.
Verdict: Both models followed the complex spatial instructions perfectly. Seedream 4.0 captures a more natural, photographic lighting setup, while Wan 2.5 (Preview) provides superior geometry and texture detail, though its sphere appears to float. Wan 2.5 is slightly preferred for adhering to 'cube' vs 'rectangular' shape.
Man and Car in California
Editing“Make a photo of the man driving the car down the California coastline”
AI Judge Analysis
Seedream 4.0
- + Excellent preservation of the man's identity, clothing, and hairstyle.
- + Seamless integration of the car and driver into a cinematic coastline environment.
- + Maintains the car's specific model details with high accuracy.
- − The driver's hands on the steering wheel look slightly merged/distorted.
- − The lighting on the man's face is a bit dark compared to the bright sunlit background.
Wan 2.5 (Preview)
- + Successfully creates a sunny California vibe with palm trees and clear blue water.
- + Dynamic motion blur on the road and wheels enhances the sense of driving.
- − Poor identity preservation; the man is barely visible and his features are lost.
- − The car's door handle and side panel details are slightly simplified compared to the source.
Verdict: Seedream 4.0 is the clear winner as it successfully composited the specific man from the source image into the car while maintaining his unique clothing and hairstyle. In contrast, Wan 2.5 (Preview) effectively changed the background but failed to preserve the man's identity, making him look like a generic figure in the distance.
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
Seedream 4.0
- + Excellent execution of car motion blur as requested
- + Detailed, realistic reflections on the wet pavement
- + Strong candid feel with the specified imperfect framing
- − The bike's mechanical structure is a bit warped, especially the chain guard and handlebars
Wan 2.5 (Preview)
- + Superb skin texture and facial detail on the subject
- + Very crisp rendering of the bicycle hardware and tools
- + Atmospheric lighting with nice bokeh from distant headlights
- − Failed to include the requested motion blur for passing cars
- − The background feel is slightly more 'polished' than 'candid street photo'
Verdict: Seedream 4.0 followed the technical cues of the prompt more accurately, specifically the motion blur and the 'imperfect' street photography aesthetic. However, Wan 2.5 (Preview) produced a significantly more life-like subject with superior skin texture and a more coherent red bicycle, despite missing the motion blur requirement.
Fantasy Warrior
Text-to-Image“Close portrait of a battle-worn paladin in ornate engraved plate armor, hair braided with small beads, faint scars and dirt on the skin, warm torchlight reflecting off metal, shallow depth of field, bokeh sparks, lifelike eyes, highly detailed texture on leather straps and cloth underlayer.”
AI Judge Analysis
Seedream 4.0
- + Excellent depiction of warm, cinematic torchlight lighting
- + Very realistic skin texture with subtle scarring
- + Highly detailed engraving on the plate armor that looks physically embossed
- − The leather straps look slightly blurry and lack some material definition
- − Braid transition into the hair at the top of the head is slightly messy
Wan 2.5 (Preview)
- + Exceptional texture on the leather straps and frayed cloth underlayer
- + Distinct beads in the braided hair as requested
- + Sharp focus on the face and intricate chainmail details
- − The lighting on the armor feels a bit flatter compared to the other model
- − The bokeh sparks in the background look like distinct orange circles rather than natural flying embers
Verdict: Both models followed the prompt closely, but Wan 2.5 (Preview) produced a sharper image with better representation of the specific 'leather and cloth' textures requested. Seedream 4.0 captures a superior cinematic mood and more realistic skin rendering, though it is slightly less sharp overall.
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
Seedream 4.0
- + Clear, readable text for category headers
- + High-quality, appetizing individual food photography
- − Lacks any actual menu content like descriptions or prices
- − Empty space in the center makes the layout feel unfinished
Wan 2.5 (Preview)
- + Successfully creates a full menu layout with descriptions and sub-headers
- + Strong adherence to the 'grid' request with a clean, professional aesthetic
- + Cohesive visual style across all food items
- − Contains significant spelling errors in both titles and body text
- − Food photos are somewhat repetitive in styling and composition
Verdict: While Seedream 4.0 provides cleaner food photography and perfectly legible headers, it fails to produce a functional menu design, leaving large gaps in the layout. Wan 2.5 (Preview) much better adheres to the prompt's structural requirements, creating a complete menu with distinct sections, vibrant accents, and a professional minimalist grid, despite the typical AI spelling artifacts.
Magic Burger Explosion: Fiery Photorealism Challenge
Text-to-Image“Ad for 'Magic Burger'. Dynamic, exploded burger with all components (bun, patty, cheese, lettuce, tomato, sauce) suspended in mid-air. Emphasize photorealistic detail and a sense of motion. Dark, fiery background with glowing embers. Integrate text: 'MAGIC BURGER' as a prominent title, 'LIMITED TIME ONLY' as a secondary message, and '€6.99' in a starburst, all rendered with a fiery, glowing effect.”
AI judge analysis unavailable for this challenge.
Chalkboard Menu
Text-to-Image“Handwritten-style chalkboard menu in a cozy café, all text rendered in the exact same realistic chalk handwriting style with natural variations in letter size, slight slant, and chalk texture — no printed or digital fonts anywhere on the board. Title at the top in elegant cursive chalk handwriting: ‘TODAY’S SPECIALS – APRIL 30, 2026’. Below it, three menu items also in the same handwritten chalk style: ‘Truffle Mushroom Risotto – $24’, ‘Grilled Octopus with Lemon & Herbs – $28’, ‘Brown Butter Chocolate Chip Cookies – $9’. At the very bottom, smaller text in the identical handwritten chalk style (slightly smaller but still clearly legible with the same handwriting characteristics): ‘All items made fresh daily • Ask about our gluten-free options’. Warm ambient café lighting, visible chalk dust and smudges, realistic handwriting imperfections, no clean printed text anywhere.”
AI Judge Analysis
Seedream 4.0
- + Excellent chalk texture including realistic dust and smudges.
- + Full prompt adherence including the completion of the 'Brown Butter' item.
- + Handwriting feels organic and varied rather than font-like.
- − The slant of the text is slightly inconsistent across lines.
Wan 2.5 (Preview)
- + Very clean and legible text rendering.
- + Good composition within the frame of the chalkboard.
- − Missed the 'Herbs' part of the octopus menu item.
- − Repeating the price '$9' on two separate lines for the cookies.
- − Text looks a bit too much like a digital font rather than natural chalk handwriting.
Verdict: Seedream 4.0 followed the prompt much more accurately, correctly completing the 'Brown Butter' item and including the 'Herbs' in the second dish. It also achieved a much more authentic chalk-on-board texture compared to Wan 2.5 (Preview), which suffered from logic errors like repeating price tags and missing prompt words.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
AI Judge Analysis
Seedream 4.0
- + Excellent texture on the astronaut suit and horse mane
- + Great lighting integration with the nebula reflection on the helmet
- + Dynamic composition with the rearing pose
- − Failed the specific spatial instruction for the horse to be on top of the astronaut
- − Anatomical issues with the horse's front legs appearing distorted
Wan 2.5 (Preview)
- + High visual clarity and vibrant cinematic colors
- + Good sense of motion and scale with the planet in the background
- + Precise rendering of the astronaut's gear
- − Failed the negative constraint/spatial instruction for the horse to be on top
- − The horse's back legs lack logical physical connection to the ground/atmosphere
Verdict: Both models failed the specific prompt instruction to place the horse on top of the astronaut, instead defaulting to the common trope of an astronaut riding a horse. Wan 2.5 (Preview) produces a cleaner, more vibrant image with better composition, while Seedream 4.0 has more realistic textures but shows significant anatomical distortion in the horse's limbs.
Outfit Transfer Challenge
Editing“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.”
AI Judge Analysis
Seedream 4.0
- + Maintains the vitiligo pattern and unique hair from Image 1 correctly.
- + Includes the specific shoes and accessories like the watch and ring.
- + Adapts the clothing to a new pose while keeping the background consistent.
- − Significantly alters the facial features of the person from Image 1, adding a mustache and changing the nose/jawline.
- − Changes the gaze of the person to look down instead of forward.
Wan 2.5 (Preview)
- + High visual quality with clean textures and realistic lighting integration.
- + Correctly applies the clothing, sunglasses, and jewelry from Image 2.
- − Fails the source preservation requirement by completely removing the vitiligo and the unique hair of the person in Image 1.
- − Replaces the person's face with the face from Image 2 instead of keeping the base person unchanged.
Verdict: Seedream 4.0 followed the specific request to keep the person from Image 1 much better than Wan 2.5, as it preserved the vitiligo and hair patterns, though it did slightly alter the facial features. Wan 2.5 (Preview) failed the core editing instruction by replacing the subject's entire identity and skin condition with the man from Image 2, essentially treating it as a background swap rather than a clothing swap.
The Capybara Taxi Driver
Text-to-Image“Photorealistic scene inside a yellow New York taxi at night. A capybara is driving, wearing a yellow taxi driver cap and a dark jacket. It has a calm, professional expression and both front paws on the steering wheel. In the back seat sits a human businesswoman in a coat, looking at her phone with a completely normal, bored expression (as if this is just another normal ride). Through the windows you can see the streets of Manhattan at night with blurred lights. Realistic taxi interior, photorealistic, detailed fur and fabric, 35mm lens, night lighting with reflections, shallow depth of field.”
AI Judge Analysis
Seedream 4.0
- + Natural composition looking through the side window
- + Accurately depicts the bored expression of the businesswoman
- + Excellent lighting integration between the city lights and the taxi interior
- − Text on the cap is nonsensical
- − Capybara paws look a bit too much like human hands
Wan 2.5 (Preview)
- + Symmetry in composition is very striking
- + High-quality fur texture and realistic rain details on the taxi
- + A more detailed and authentic-looking driver's cap
- − View of the backseat person is slightly obstructed and less natural
- − The capybara's paws are somewhat small and awkwardly positioned on the wheel
Verdict: Both models followed the prompt excellently, capturing the surreal yet professional atmosphere. Seedream 4.0 is preferred for its more natural, candid composition and the perfect execution of the businesswoman's bored expression, whereas Wan 2.5 (Preview) excels in texture and environmental details like the rain and vibrant city lights.
The Halloween Invitation
Text-to-Image“Vintage gothic Halloween party invitation. Dark parchment poster, spooky border with webs and thorns, central glowing jack-o-lantern, bats, twisted trees, moody night sky. Add elegant gothic title text saying "Halloween Party Invitation", a small scroll banner saying "You are invited to a night of frights", and event details at the bottom: Date: 30.10.2026 Time: 7pm Location: The Arches, NYC Spooky but polished, cinematic lighting, square format.”
AI Judge Analysis
Seedream 4.0
- + Excellent moody atmosphere and cinematic lighting consistent with the 'vintage' prompt.
- + Text is mostly integrated well into the scene.
- + The thistly thorn border feels more organic and integrated with the background.
- − The text in the small scroll banner is garbled and barely legible.
- − The bottom text has minor alignment issues and font inconsistencies.
Wan 2.5 (Preview)
- + Superior text rendering with perfect legibility and zero spelling errors.
- + The scroll banner looks very polished and physically realistic.
- + Higher resolution detail on the jack-o-lantern and background elements.
- − The composition feels slightly more like a modern graphic design than a vintage parchment poster.
- − The transition between the blue night sky circle and the parchment is very sharp and less 'painterly'.
Verdict: Both models followed the prompt's structural requirements perfectly. Seedream 4.0 captures a better 'vintage gothic' mood with its dark, atmospheric lighting, but suffers from poor text rendering on the scroll. Wan 2.5 (Preview) is the clear winner for an actual invitation because its text is perfectly legible and professional, even though the overall aesthetic feels slightly more modern.
Bald man challenge
Image 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
Seedream 4.0
- + Excellent preservation of the original facial features
- + Consistent lighting across the new hair and existing face
- − The hair has a stiff, fuzzy texture that looks like a wig
- − The hairline is very blocky and lacks natural transitions
Wan 2.5 (Preview)
- + Realistic hair texture with natural flow and highlights
- + Integration with sideburns and temples is much more seamless
- + Hairline looks convincing and non-linear
- − Slightly altered the shape of the nose and the squint of the eyes compared to the original
Verdict: Seedream 4.0 does a perfect job of preserving the original face, but the added hair looks like a low-quality wig with an artificial texture. Wan 2.5 (Preview) produces a significantly more realistic and stylish head of hair that integrates naturally with the subject, even though it slightly modifies the facial anatomy in the process.
Isometric Miniature Diorama Scenes
Text-to-Image“Create a clear, 45° top-down isometric miniature 3D cartoon scene of Japan's signature dish: sushi, with soft refined textures, realistic PBR materials, gentle lighting, on a small raised diorama base with minimal garnish and plate. Solid light blue background. At top-center: 'JAPAN' in large bold text, 'SUSHI' below it, small flag icon. Perfectly centered, ultra-clean, high-clarity, square format.”
AI Judge Analysis
Seedream 4.0
- + Excellent variety of sushi types following the 'miniature scene' request.
- + Perfect text alignment and placement of the Japan flag icon.
- + Higher level of detail in the food textures, such as the rice grains and fish gloss.
- − The 'JAPAN' text is simple black rather than stylized integrated 3D text.
- − The background is slightly darker than requested.
Wan 2.5 (Preview)
- + Beautiful 3D typography that fits the 'cartoon' and 'PBR' aesthetic well.
- + Very clean, minimal diorama base as requested by the prompt.
- + Soft, pleasing lighting and smooth textures.
- − Lower variety of sushi, showing only one piece instead of a 'scene'.
- − The perspective is more eye-level than the requested 45° top-down isometric view.
Verdict: Seedream 4.0 followed the prompt's layout and variety requirements more closely, providing a comprehensive sushi platter at the correct isometric angle. While Wan 2.5 (Preview) produced more visually appealing 3D text and a cleaner professional render, it failed to provide a 'scene' of sushi and ignored the specific angle instruction.
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
Seedream 4.0
- + Excellent preservation of the subject's facial features while applying caricature proportions.
- + Retains the original background and clothing style for better consistency.
- + Creative integration of the profession through a news desk and microphone.
- − The anatomy of the right hand holding the phone is slightly mangled.
- − The hockey integration is a bit subtler on the desk panel compared to the other themes.
Wan 2.5 (Preview)
- + Strong cartoon illustration style that fits the 'exaggerated' part of the prompt.
- + Clear and various nods to all elements: hockey on the screen, sticks, and jerseys.
- + Symmetrical and balanced composition.
- − Loses the specific likeness of the subject, appearing more like a generic clipart character.
- − Completely ignores the source image background and lighting.
- − The hand holding the microphone is poorly rendered with merged fingers.
Verdict: Seedream 4.0 is the superior choice for an image editing task as it successfully applies the caricature effect while maintaining the subject's distinct identity and the original photo's environment. Wan 2.5 (Preview) ignores the source image's context and creates a generic digital illustration that lacks the subject's actual likeness.
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
Seedream 4.0
- + Natural and dynamic posing of the animals tumbling and playing.
- + Beautiful rendering of backlighting and translucent fur edges.
- + Excellent integration of the creatures within the environment's depth of field.
- − The fox's anatomy, specifically its front legs and paws, appears slightly distorted.
- − A high amount of bokeh particles can look a bit cluttered in the upper half.
Wan 2.5 (Preview)
- + Clearer full-body rendering of all four animals from a front-facing perspective.
- + Vibrant color palette and sharp detail on the flower foreground.
- + Clean and bright lighting that matches the 'joyful' vibe of the prompt.
- − The fox has unsettling, overly human-like or artificial glowing blue eyes.
- − The floating water droplets look like stickers rather than realistic dew.
- − The animals don't appear to be interacting with each other as much as just standing in a row.
Verdict: Seedream 4.0 captures a more artistic and believable scene with dynamic movement and natural lighting, feeling like a single cohesive moment. In contrast, Wan 2.5 (Preview) presents the animals in a somewhat static, 'family portrait' style with several jarring artifacts, particularly the fox's eyes and the physics-defying dew drops. Seedream 4.0 is the preferred choice for its superior composition and realistic fur rendering.
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
Seedream 4.0
- + Successfully captures a hand-painted watercolor texture similar to 'The Tale of the Princess Kaguya'.
- + Executes the soft pastel color palette and dreamy background perfectly.
- + Preserves the iconic composition and poses with high fidelity.
- − The faces are a bit simplified, losing some of the specific expression nuances from the original meme.
Wan 2.5 (Preview)
- + Accurately translates the original facial expressions into a classic 90s/00s Ghibli character design style.
- + Preserves background details from the original photo while stylizing them.
- + Creatively adds falling leaves to enhance the 'nostalgic' mood.
- − The lighting is a bit flat compared to the requested 'gentle, dreamy' atmosphere.
- − The texture feels more like digital cell-shading than the requested 'hand-painted' texture.
Verdict: Both models did an excellent job of preserving the structure of the original 'distracted boyfriend' meme while applying the Ghibli theme. Seedream 4.0 is preferred for its superior adherence to the artistic medium requested (hand-painted textures and soft pastels), whereas Wan 2.5 (Preview) leans more toward a modern digital anime aesthetic.
Golden Hour Stroll
Image Editing“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
AI Judge Analysis
Seedream 4.0
- + Excellent depiction of wind-blown hair that looks natural
- + Added autumnal leaves that match the colors of the background foliage
- + Preserved the identity of the person and dog very accurately
- − Introduction of a lens flare in the top right that wasn't requested
- − The leaves have a slightly blurry, low-resolution appearance compared to the subjects
Wan 2.5 (Preview)
- + Successfully applied the hair motion effect while maintaining high sharpness
- + Preserved the original lighting and colors of the source image perfectly
- + Maintained the facial structure and dog details without distortion
- − The added leaves look like flat, green 2D stickers that don't match the environment's lighting or color
- − Less 'dynamic' feel in the leaves compared to Model A
Verdict: Both models followed the core instructions well, but Seedream 4.0 created a more cohesive environment by adding warm, autumnal leaves that blended with the background, despite adding an unrequested lens flare. Wan 2.5 (Preview) produced a higher quality hair effect but failed on the leaf integration, as they appear as bright green, flat shapes that distract from the realism of the scene.
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
Seedream 4.0
- + Excellent typography and spelling of 'Caffè Florian'
- + Clean and balanced composition following use of white space
- + Good vector-style shading on the cloche
- − The steam effect is a bit stiff compared to the rest of the logo
Wan 2.5 (Preview)
- + Very cohesive emblem layout within the banner
- + Elegant steam illustration
- + High-quality vintage paper texture effect
- − The accents on the 'e' in 'Caffè' are messy and inconsistent
- − The small 'Est. 1720' text is slightly off-center
Verdict: Both models followed the prompt closely, providing high-quality logos with correct vintage aesthetics. Seedream 4.0 is slightly better due to its cleaner font rendering and better adherence to the specific accent mark in 'Caffè', whereas Wan 2.5 (Preview) struggles with the clarity of the typography on the ribbon.
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
Seedream 4.0
- + Strong flat-vector aesthetic that matches the requested style perfectly.
- + Good spatial layout separating the sequence of events from the landing surface.
- + Includes all six requested steps clearly numbered.
- − Contains spelling errors like 'Surfcce' and 'Translunar' spelled as 'TransLunar'.
- − The rocket for the descent stage looks more like a modern capsule/shuttle than a Lunar Module.
Wan 2.5 (Preview)
- + Excellent iconography for the Saturn V and the Lunar Module.
- + Clean, professional typography without spelling errors.
- + Sophisticated use of the NASA-inspired color palette.
- − Step numbering is missing, making the sequence harder to follow.
- − The 'Translunar' step includes a second large moon which is confusing in a flat-vector infographic.
- − Translunar trajectory line intersects other elements haphazardly.
Verdict: Seedream 4.0 followed the specific structural instructions and numbering better, but suffered from minor typos and less accurate spacecraft silhouettes. Wan 2.5 (Preview) produced a higher quality, more professional graphic with much better technical details in the iconography, despite missing the specific step-by-step numbering requested.
Explore each model
Alibaba's text-to-image and image-to-image generation model from the Wan AI suite, offering high-quality visual generation capabilities