FP8 quantized variant of Black Forest Labs' FLUX.1 [schnell] model, offering ~2x faster inference with reduced precision while maintaining high-quality image generation in 4 steps
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.1 [schnell] FP8
19.0
arena score
#46 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Seedream 4.0
24.7
arena score
#15 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.1 [schnell] FP8
0%
win rate
Ties
0%
Seedream 4.0
0%
win rate
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent photorealism and crisp textures
- + Beautiful high-end product photography aesthetic
- + Accurate soft lighting and depth of field
- − The glass container is more of a tall rectangular prism than a cube
- − The blue globe is floating on a strange internal shelf rather than resting on the bottom
Seedream 4.0
- + Accurately depicts the spatial relationships between the objects
- + The container is a perfect cube as requested
- + Successfully shows the plant behind the cube visible through the glass
- − The glass texture looks slightly dirty or low-resolution compared to Model A
- − The shadows on the table are a bit harsh and busy compared to the 'soft window light' prompt
Verdict: Both models followed the prompt successfully, including all four objects and specific lighting. Model A (FLUX.1 [schnell] FP8) produced a much more visually stunning and realistic image, though it failed the 'cube' instruction by creating a prism; Seedream 4.0 followed the geometric and spatial instructions more precisely but with lower render quality.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent puddles and wet pavement reflections.
- + Vibrant color palette and cinematic lighting.
- + High skin texture detail in the subject.
- − Fails to capture the motion blur requested.
- − The man is just holding the handlebars rather than repairing the bike.
- − Lacks the visible light rain effect.
Seedream 4.0
- + Follows the 'repairing' prompt accurately with the man seated and tools out.
- + Correctly implements motion blur on the passing vehicle.
- + Includes visible rain droplets and a more authentic 'imperfect' frame.
- − The bike anatomy is slightly warped, especially the chain guard area.
- − Lower overall sharpness compared to model A.
- − The reflection in the puddle is somewhat muddy and less defined.
Verdict: Seedream 4.0 is the superior image for this specific prompt as it correctly interprets 'repairing' and 'motion blur', whereas FLUX.1 [schnell] FP8 provides a static, clean portrait of a man holding a bike. Seedream 4.0 better captures the gritty, candid atmosphere and technical camera effects requested, despite its slightly lower resolution.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Strong facial expression and high-contrast dramatic lighting
- + Excellent engraving details on the armor and neck gorget
- + Very sharp and clean skin texture
- − Missed the request for braided hair with beads
- − The character looks more like a rogue or ranger than a battle-worn paladin
- − Lacks visible sparks requested in the prompt
Seedream 4.0
- + Perfect adherence to braided hair with beads and visible scars/dirt
- + Includes all requested atmospheric elements like the torch and bokeh sparks
- + Excellent rendering of the leather straps and underlayer textures
- − Slightly softer facial focus compared to model A
- − Minor lighting inconsistency on the left side of the face relative to the torch position
Verdict: While FLUX.1 [schnell] FP8 offers a higher level of raw sharpness and detail in the engraving, Seedream 4.0 followed the specific prompt instructions much more accurately, including the braided hair with beads, scars, and bokeh sparks. Seedream 4.0 better captures the 'battle-worn paladin' archetype and the environmental storytelling requested.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent structure that resembles a real, multi-page restaurant menu.
- + Clear sections for Appetizers, Pizza, and Mains as requested.
- + Realistic alignment of text and prices following a professional layout.
- − Nonsense 'lorem ipsum' style text and spelling errors like 'APPTIZERS' and 'PIZZAL'.
- − The header text 'ORCETERS' and 'SECCER' is gibberish.
Seedream 4.0
- + Clean, high-quality photography with vibrant colors.
- + Large, bold sans-serif header fonts that are perfectly legible.
- + Effective use of grid-based design for food images.
- − Fails to include actual menu text, descriptions, or prices.
- − Layout looks more like a mood board or cover page than a functional menu.
Verdict: FLUX.1 [schnell] FP8 successfully captures the complex structure of a restaurant menu with prices and item lists, though it suffers from significant spelling errors and gibberish text. Seedream 4.0 produces beautiful imagery and clean typography, but it lacks the functional elements required for a full menu design. FLUX.1 [schnell] FP8 is the better choice for this specific prompt as it adheres more closely to the 'sections' and 'layout' requirements.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Features a very high-quality, vibrant food render.
- + Good clarity on the secondary ingredients like floating cheese cubes and tomatoes.
- − Serious spelling errors including 'LIIMITED' and 'NEEY'.
- − The burger is not truly 'exploded' but mostly intact with small bits floating around it.
- − The starburst and price are plain and do not follow the fiery theme requested.
Seedream 4.0
- + Excellent adherence to the 'exploded' and 'dynamic motion' request with swirling effects.
- + Perfect text rendering with the requested fiery, glowing effect.
- + Price is contained in a fiery starburst as prompted, although the digit is slightly off.
- − The price is rendered as €5.99 instead of the requested €6.99.
- − The bottom half of the burger is duplicated/doubled up rather than separated into individual components.
Verdict: Seedream 4.0 is the clear winner because it successfully followed the stylistic instructions for fiery, glowing text and a dynamic exploded composition. FLUX.1 [schnell] FP8 failed significantly on the text rendering, producing multiple spelling errors and a flat, unintegrated starburst sticker.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent wood frame and lighting on the board surface.
- + Legible date rendering.
- − Poor text accuracy with numerous spelling errors and repeated lines.
- − Non-cursive text for the title contrary to the prompt.
- − The 'chalk' texture looks more like a white ink marker.
Seedream 4.0
- + Perfect adherence to text content with zero spelling errors.
- + Authentic chalk texture and smudges for a realistic chalkboard feel.
- + Followed the directive for elegant cursive handwriting in the title.
- − The perspective on the bottom of the board is slightly warped.
Verdict: Seedream 4.0 is much better due to its flawless rendering of the requested text and superior adherence to stylistic instructions like cursive handwriting and chalk texture. FLUX.1 [schnell] FP8 struggled significantly with text coherence, resulting in numerous repetitions and gibberish words.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.1 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Successfully followed the difficult logic instruction of the horse on top of the astronaut/equipment.
- + High-quality textures on the horse's fur and mechanical components.
- + Strong cinematic composition with the Earth as a backdrop.
- − The astronaut is represented more as a mechanical suit or box rather than a clear human figure.
- − Anatomically awkward second horse head in the background.
Seedream 4.0
- + Excellent visual clarity and vibrant colors in the nebula background.
- + Very high detail in the astronaut suit and horse mane.
- − Failed the core prompt instruction of placing the horse on top of the astronaut.
- − Anatomical issues with the astronaut's leg placement and horse's front legs.
Verdict: FLUX.1 [schnell] FP8 is the clear winner because it understood the inverted logic of the prompt, placing the horse on top of the astronaut/equipment. Seedream 4.0 produced a high-quality but generic image of an astronaut riding a horse, which directly ignored the specific 'not vice versa' instruction.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent text rendering on the hat
- + Superior lighting and color saturation
- + High level of detail in the capybara's fur and the female subject
- − The passenger is holding two phones simultaneously
- − The capybara's paws are not quite on the steering wheel correctly
Seedream 4.0
- + Accurately depicts both paws on the steering wheel
- + Realistic depth of field and authentic taxi interior perspective
- + Better captures the 'bored' expression of the passenger
- − Text on the hat is corrupted ('TAT' instead of 'TAXI')
- − Overall image is slightly grainier compared to the other model
Verdict: FLUX.1 [schnell] FP8 produces a more vibrant and aesthetically pleasing image with perfect text rendering, but fails on internal logic by giving the passenger two phones. Seedream 4.0 follows the physical instructions of the prompt more accurately (paws on wheel, bored expression) but suffers from text artifacts and lower resolution.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Strong cinematic lighting on the central jack-o-lantern.
- + Clean, legible font choice for the main heading.
- − Numerous spelling errors in the banner and event details (e.g., 'A ai nigh tof friights', 'Tine').
- − Border lacks the requested webs and thorns, appearing more like plain parchment edges.
- − Composition feels slightly empty in the bottom half.
Seedream 4.0
- + Excellent adherence to all visual prompts, including thorny borders and spiderwebs.
- + Text accuracy is high, correctly rendering the date, time, and location as requested.
- + Atmospheric composition with realistic textures on the trees and pumpkin.
- − Minor text distortion on the smaller scroll banner ('invitcd').
- − The border thorns appear somewhat repetitive and overlay the parchment harshly in the corners.
Verdict: Seedream 4.0 followed the prompt much more accurately, including all specific decorative elements like the thorns and webs while maintaining high text accuracy for the event details. FLUX.1 [schnell] FP8 struggled significantly with text rendering and missed several key atmospheric prompts, resulting in a generic layout with many typos.
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 [schnell] FP8
Seedream 4.0
AI judge analysis unavailable for this challenge.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Strong vertical composition and vibrant colors
- + Captures very expressive facial features on the animals
- + Achieves soft, glowing lighting effects
- − Failed to include the requested baby bunny
- − Anatomically incorrect extra animals that appear to be fox-cat hybrids
- − More illustrative/digital art style than hyper-photorealistic
Seedream 4.0
- + Successfully includes all four requested animals: puppy, kitten, bunny, and fox
- + Excellent adherence to the 'playfully chasing' and 'tumbling' prompt with dynamic posing
- + Highly realistic rendering of dew sparkles and golden god rays
- − The fox kit has slightly strange dark paws/arms
- − One butterfly has a somewhat messy wing structure
Verdict: Seedream 4.0 is the clear winner as it followed all prompt instructions, including the specific list of animals and the active 'tumbling' motion. FLUX.1 [schnell] FP8 failed to include the bunny and generated several indistinguishable hybrid animals, resulting in a less coherent scene despite the pleasing lighting.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Strong decorative vector style
- + Good execution of the warm brown and cream color scheme
- + Captures the architectural dome feel of the real Caffè Florian
- − Serious spelling errors in the brand name ('AFe FLAMILAN')
- − The steam element is stylized to the point of being unrecognizable
- − Muddled banner composition
Seedream 4.0
- + Perfect text rendering for both specific phrases
- + Clear and literal interpretation of the 'cloche dome and steam' request
- + Excellent subtle paper texture on the background
- − Composition feels a bit fragmented with the arched text far from the emblem
- − Simplified cloche design looks more like a dinner dish than a high-end cafe logo
Verdict: Seedream 4.0 is the clear winner because it followed all text prompts accurately, whereas FLUX.1 [schnell] FP8 hallucinated the brand name significantly. Seedream 4.0 also provided the requested 'subtle texture' and 'steam' elements with much more clarity and professional layout.
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 [schnell] FP8
Seedream 4.0
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Strong minimalist vector aesthetic
- + Excellent use of the specified NASA-inspired color palette
- + Professional layout structure resembling a real technical infographic
- − Nonsense filler text and widespread typos
- − Icons are abstract and do not clearly represent the specific requested objects like the Saturn V
Seedream 4.0
- + Successfully followed the specific icon requests for all 6 steps
- + Text is largely legible and names the astronauts correctly
- + Good balance of technical diagrams and illustrative elements
- − Text is slightly uneven with minor spelling errors like 'Surfece'
- − Visual style is a bit more 'clip-art' than 'modern vector infographic'
Verdict: Both models captured the essence of the prompt, but Seedream 4.0 is the clear winner for its superior prompt adherence regarding the specific 6-step sequence. While FLUX.1 [schnell] FP8 has a more sophisticated aesthetic and better color handling, it failed to provide recognizable icons and rendered mostly gibberish text, whereas Seedream 4.0 provided accurate labels and distinct, appropriate imagery for each stage of the mission.
Next steps
Explore each model
ByteDance's image generation model with integrated text-to-image and image editing capabilities in a unified architecture, supporting up to 4K resolution