Black Forest Labs' 12-billion parameter flow transformer for high-quality text-to-image generation, suitable for personal and commercial use with streaming support
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.1 [dev]
24.6
arena score
#16 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Seedream 4.5
26.1
arena score
#9 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.1 [dev]
100.0%
win rate
Ties
0.0%
Seedream 4.5
0.0%
win rate
100.0%
0.0% ties
0.0%
Shared challenges
13
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent photographic quality and sharp textures on the book leather.
- + Sophisticated lighting and shallow depth of field.
- + Accurate spatial arrangement of objects.
- − The glass 'cube' has complex internal structures that make it look more like a glass block frame than a simple hollow cube.
- − The blue sphere appears to be floating mid-air inside the glass.
Seedream 4.5
- + Perfect adherence to the object requirements, specifically the hollow glass cube.
- + Exceptional realism in reflections and caustic light on the table.
- + The scale of the 'small blue sphere' is more accurate to the prompt.
- − The plant is slightly more blurred/abstract in the background compared to the clear leaves in the other version.
- − The composition is a bit tighter and more centered.
Verdict: Both models followed the prompt perfectly, but Seedream 4.5 is the winner due to its superior handling of physical realism, particularly how the light passes through the glass cube to create realistic shadows and reflections on the table. FLUX.1 [dev] produced a beautiful image, but its glass cube is structurally confusing and the sphere appears to be magically floating, whereas Seedream 4.5 created a more believable scene.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent shallow depth of field and bokeh realism
- + Natural skin textures and subtle facial details
- + Accurate 50mm lens perspective
- − Lack of motion blur on the passing car as requested
- − The man appears to be simply holding the bike rather than repairing it
- − Raindrops look like static noise rather than falling streaks
Seedream 4.5
- + Successfully incorporates motion blur from passing cars
- + Stronger narrative with the man actually crouching to repair the chain
- + Captures the 'imperfect framing' and 'candid' feel much better
- − Slightly over-saturated neon reflections feel less 'no stylization'
- − The rain droplets on the jacket are a bit oversized and uniform
- − The man's hands have slightly messy anatomical details near the bike chain
Verdict: Seedream 4.5 is the winner as it followed more of the complex prompt requirements, specifically the motion blur and the actual action of 'repairing' the bicycle. While FLUX.1 [dev] produced a cleaner portrait with better skin textures, it missed the dynamic elements that make the scene feel like a candid street photo.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent photographic quality and lighting on the skin
- + Subtle but effective bokeh sparks in the background
- − Failed to include 'hair braided with small beads' (braids are present, beads are not)
- − Armor lacks 'ornate engraving' and looks relatively plain
- − Skin looks too clean and youthful, failing the 'battle-worn' and 'scars' prompt
Seedream 4.5
- + Perfect adherence to specific details like small silver beads in the hair
- + Exquisite texture on the ornate armor, leather straps, and cloth under-layers
- + Realistically captures the 'battle-worn' aesthetic with dirt and visible scars
- − The fire on the left is a bit blurry compared to the rest of the high-fidelity foreground
Verdict: Seedream 4.5 followed the prompt much more accurately, capturing every specific detail from the beaded braids to the ornate engravings and the battle-worn skin. FLUX.1 [dev] produced a high-quality portrait but failed several key descriptors, presenting a character that looked more like a clean model than a scarred paladin.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent professional layout that feels like a real restaurant menu
- + Clean and readable sans-serif typography with coherent hierarchy
- + Food images are integrated tastefully with minimalist white space
- − Failed to include a dedicated 'Pizza' section as requested in the categories
- − Food photos are scattered rather than in the requested 'grid' layout
Seedream 4.5
- + Perfect adherence to the 'grid' structure requested
- + Includes all three requested sections: Appetizers, Pizza, and Mains
- + Vibrant color-coded accents around the food photos
- − Text is repetitive with 'Pizza' and 'Restaurant' used as line items
- − Layout feels more like a template preview than a finished design
- − Composition is a bit cramped with high price points that clash with 'casual dining'
Verdict: Seedream 4.5 followed the structural prompts much better by providing the specific grid layout and all three requested sections (Appetizers/Pizza/Mains). However, FLUX.1 [dev] produced a far more professional and aesthetically pleasing design that looks like an actual menu, despite missing the pizza category. Seedream 4.5 is the winner for strict prompt adherence, while FLUX.1 [dev] is superior in visual quality and realism.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Crisp, high-resolution textures on the ingredients.
- + Excellent vertical composition showing clearly separated layers.
- − Failed to include the primary 'MAGIC BURGER' title text.
- − Missing the required starburst for the price element.
- − Background is quite static and lacks the 'fiery' energy requested.
Seedream 4.5
- + Perfect adherence to all text requirements including the 'MAGIC BURGER' title and starburst.
- + Dynamic sense of motion with motion-blurred ingredients and embers.
- + Strong fiery aesthetic that matches the 'dark, fiery background' prompt.
- − Text in the starburst is slightly less crisp than the main title.
- − The 'exploded' effect is less literal than Model A, though more dynamic.
Verdict: While FLUX.1 [dev] produced a very clean and appetizing image, it failed to follow the core text instructions of the prompt, omitting the brand name entirely. Seedream 4.5 followed every instruction, including the complex text rendering and the specific starburst price tag, while creating a much more energetic and atmospheric advertisement.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent text legibility and spelling accuracy.
- + Good center alignment and spacing of elements.
- + Vibrant and clean presentation.
- − The 'chalk' texture looks too clean and more like a digital marker or vinyl decal than actual chalk dust.
- − Includes minor spelling artifacts in the bottom disclaimer ('drish' instead of 'dish').
Seedream 4.5
- + Superb adherence to the 'chalk' texture requirement, with realistic smudges and graininess.
- + Accurately represents the slant and natural variation of real handwriting.
- + Spells all text correctly, including the complex menu items.
- − The pricing for the first item is repetitive, listing '$24' twice on two lines.
- − The composition is a bit more crowded than Model A.
Verdict: While FLUX.1 [dev] produced a cleaner and more legible sign, Seedream 4.5 far surpassed it in terms of prompt adherence regarding the texture and feel of real chalk. Seedream 4.5 captured the requested natural variations and chalky residue that make the image look authentic, whereas FLUX.1 [dev] looked too much like a digital font despite the instructions.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + High visual clarity and cinematic lighting.
- + Clean composition with a sleek sci-fi aesthetic.
- − Completely failed the negative constraint/spatial instruction of 'horse on top'.
- − Anatomical issues with the horse's back legs.
Seedream 4.5
- + Beautiful color palette with vibrant nebula effects.
- + Detailed textures on the horse's fur and the astronaut's suit.
- − Completely failed the negative constraint/spatial instruction of 'horse on top'.
- − Some anatomical confusion where the horse's legs and the astronaut's legs overlap.
Verdict: Both FLUX.1 [dev] and Seedream 4.5 failed the specific negative constraint to place the 'horse on top' of the astronaut, instead defaulting to the common trope of an astronaut riding a horse. FLUX.1 [dev] offers a cleaner, more minimalist cinematic look, while Seedream 4.5 provides a more vibrant and detailed space background, but neither followed the core inversion requested in the prompt.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent fur texture and lighting on the capybara.
- + Successfully captured a professional, focused capybara expression.
- + High cinematic quality with smooth bokeh and realistic interior lighting.
- − Failed the spatial request; the passenger is sitting in the front seat beside the driver instead of the back seat.
- − The passenger is looking at the capybara rather than looking down at her phone with a bored expression.
- − The perspective feels slightly claustrophobic compared to a standard taxi view.
Seedream 4.5
- + Perfect adherence to the spatial prompt, with the businesswoman correctly positioned in the back seat.
- + Accurately depicted the passenger's 'bored' expression and focus on her phone.
- + Both paws are clearly placed on the steering wheel as requested.
- − The passenger's hands and phone have some minor anatomical distortions/blurring.
- − The exterior lighting reflected on the car glass is a bit harsh and less integrated than model A.
Verdict: While FLUX.1 [dev] produced a more aesthetically pleasing and high-resolution image of the capybara itself, it failed significantly on the composition by placing the passenger in the front seat. Seedream 4.5 followed every instruction in the prompt perfectly, including the boring/everyday atmosphere and the specific seating arrangement, making it the better conceptual match despite slightly lower fine-detail quality in the human's hands.
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.”
FLUX.1 [dev]
Seedream 4.5
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.1 [dev]
- + Features a cohesive thorn border as requested.
- + Strong artistic illustration style with vibrant glowing lighting.
- − Significant spelling errors in the main title.
- − Unnecessary repetition of the time and typos in the location text.
Seedream 4.5
- + Excellent text rendering with accurate spelling throughout.
- + Effective cinematic lighting and atmospheric foggy background.
- + Correctly incorporates all requested elements including webs, thorns, and banner.
- − The thorns are positioned at the bottom rather than framing the image as a border.
- − The layout is a bit top-heavy with the text placement.
Verdict: Seedream 4.5 delivers a superior product for an invitation due to its near-perfect text rendering and atmospheric cinematic quality. While FLUX.1 [dev] has a nice illustrative style, its failure to spell the primary title and event details correctly makes it less functional.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent soft 3D cartoon textures and clay-like aesthetics
- + Accurate miniature diorama style with elegant framing
- − Text rendering is poor with spelling errors like 'SUSH CATON'
- − Japanese flag icon is placed awkwardly inside the text block
Seedream 4.5
- + Perfect text rendering for both 'JAPAN' and 'SUSHI'
- + Excellent materials that match the 'realistic PBR' and 'soft' request
- + Logical placement of the small flag icon
- − Shadows are a bit harsh for 'gentle lighting'
- − Slightly less 'isometric' tilt compared to Image A
Verdict: While FLUX.1 [dev] captures a more artistic and miniature diorama vibe, it fails significantly on the text rendering. Seedream 4.5 is the clear winner for following every part of the prompt, including perfect text, high-quality PBR materials, and a clean layout that feels professional and intentional.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent character consistency in a stylized, 3D-animation aesthetic.
- + Very clean composition with no anatomical glitches.
- + Bright, cheerful lighting that fits the 'joyful' prompt requirement.
- − Failed multiple prompt requirements: no kitten and no fox (appears to be two puppies and two rabbits/fox-hybrids).
- − Completely missed the 'hyper-photorealistic' instruction, opting for a cartoonish style.
Seedream 4.5
- + Successfully included all four distinct species: golden retriever, tabby kitten, bunny, and red fox.
- + High adherence to the 'hyper-photorealistic' style with detailed fur and realistic lighting.
- + Dynamic composition with god rays and dew sparkles as requested.
- − The fox has some anatomical issues with its right-front paw.
- − Slightly cluttered foreground with bokeh elements that can look like artifacts.
Verdict: Seedream 4.5 is the clear winner as it followed all prompt instructions, including the specific list of four different animal species which FLUX.1 [dev] failed to distinguish. While Seedream 4.5 has a minor anatomical glitch on the fox, its hyper-photorealistic rendering and inclusion of the kitten and fox make it far more accurate than the stylized, incorrect output from FLUX.1 [dev].
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Clean vector aesthetic with good use of negative space
- + Excellent decorative banner and emblem style
- + Accurate color palette according to the prompt
- − Several spelling errors including 'Flariláan' and 'Reseaurant'
- − Includes conflicting dates of 11011 and 1941 which were not requested
Seedream 4.5
- + Perfect text rendering for 'Caffè Florian' and 'Est. 1720'
- + Clean minimalist composition that adheres well to a vintage logo style
- + Correct implementation of the cloche dome with steam
- − The 'Est. 1720' font is a bit standard compared to the primary logo font
- − The texture on the background is very subtle, almost unnoticeable
Verdict: Seedream 4.5 is the clear winner because it correctly spelled all requested text, whereas FLUX.1 [dev] produced significant typos and added random dates that cluttered the design. Seedream 4.5 also better captured the clean, minimalist vector emblem look requested in the prompt.
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.”
FLUX.1 [dev]
Seedream 4.5
AI Judge Analysis
FLUX.1 [dev]
- + Excellent flat-vector aesthetic with a cohesive artistic style.
- + Captures the requested color palette well with sophisticated muted tones.
- + Good use of layout and spacing for a poster design.
- − Text consists of nonsensical gibberish instead of the requested labels.
- − Fails to follow the logical 6-step progression requested in the prompt.
- − Iconography includes irrelevant elements like ringed planets not related to the moon.
Seedream 4.5
- + Perfect adherence to the 6-step chronological structure requested.
- + Excellent text rendering for all step labels and crew names.
- + Accurate iconography for the Saturn V and Lunar Module.
- − The color palette is a bit harsh compared to the 'muted' request.
- − Layout feels slightly cramped with the large horizontal divider.
- − Includes an odd satellite icon for 'Descent' that doesn't quite match the action.
Verdict: While FLUX.1 [dev] produced a more stylish and visually attractive vector illustration, it failed significantly on the informational requirements and text accuracy. Seedream 4.5 followed the complex 6-step instructions perfectly and rendered all text legibly, making it a much more successful infographic overall.
Next steps
Explore each model
ByteDance's latest image generation model unifying text-to-image and image editing in a single architecture, with improved text rendering and 30-40% faster generation than v4.0