Black Forest Labs' aesthetically-tuned 12-billion parameter flow transformer optimized for high-quality images with incredible aesthetics, suitable for personal and commercial use
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.1 Krea [dev]
18.8
arena score
#48 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Stable Diffusion 3.5 Large Turbo
10.7
arena score
#61 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.1 Krea [dev]
0%
win rate
Ties
0%
Stable Diffusion 3.5 Large Turbo
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Perfectly follows the spatial requirements including the difficult 'behind the cube' placement.
- + Exceptional realism in materials, particularly the glass refractions and wood texture.
- + The plant is realistically visible through several layers of glass.
- − The blue sphere has a slight internal reflection that looks a bit busy.
Stable Diffusion 3.5 Large Turbo
- + Clean, high-contrast lighting that emphasizes the geometry.
- + Vibrant colors and sharp edges on the objects.
- − Failed spatial reasoning by placing the red book inside the cube instead of on top.
- − The plant is not visible through the glass as requested; it is only to the left of the cube.
- − The cube looks more like a frame than a solid glass object.
Verdict: FLUX.1 Krea [dev] followed the complex spatial instructions perfectly, placing every object in its requested location with highly realistic material rendering. Stable Diffusion 3.5 Large Turbo failed the prompt engineering by placing the red book inside rather than on top of the cube, and missed the requirement to have the plant visible through the glass.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent adherence to the 'motion blur from passing cars' prompt requirement.
- + Realistic cinematic lighting and convincing wet pavement reflections.
- + Very realistic skin and hair texture on the subject.
- − The character is standing with the bike rather than actively 'repairing' it as requested.
- − The bicycle's physical geometry is slightly warped near the pedal and chain area.
Stable Diffusion 3.5 Large Turbo
- + Successfully depicts the 'repairing' action with the subject leaning over the handlebars.
- + The bicycle structure is generally clean and consistent for a Turbo model.
- − Failed to include motion blur for the passing cars, which appear static.
- − The subject's head and hat are rendered with significant anatomical distortions and smudging.
- − The overall image quality lacks the requested realism and feels more like a 3D render.
Verdict: FLUX.1 Krea (dev) significantly outperformed the competitor by capturing the specific atmospheric elements requested, such as motion blur and natural skin textures, resulting in a believable street photograph. Stable Diffusion 3.5 Large Turbo struggled with human anatomy and failed to execute the motion blur, though it did better at posing the character in a 'repairing' stance.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent adherence to the 'battle-worn' aesthetic with realistic grime and skin textures.
- + Superior engraving detail on the plate armor and realistic fabric textures.
- + Highly effective use of lighting and bokeh to create a cinematic atmosphere.
- − The beads in the hair are relatively small and subtle compared to the request.
Stable Diffusion 3.5 Large Turbo
- + Clear braids and thick hair texture.
- + High contrast and sharp details on the armor studs.
- − The skin texture appears overly smooth and plastic, lacking the 'lifelike' quality requested.
- − The 'blood' or dirt on the face looks like flat digital paint rather than realistic wounds.
- − Anatomical issues with the ear and how the hair integrates into the head.
Verdict: FLUX.1 Krea [dev] significantly outperforms Stable Diffusion 3.5 Large Turbo by delivering a gritty, realistic, and anatomically correct portrait that perfectly captures the 'battle-worn' prompt. Stable Diffusion 3.5 Large Turbo produces a more stylized, almost AI-typical 'plastic' finish with less convincing textures on the skin and armor.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent professional layout that looks like a real-world restaurant flyer.
- + High-quality, realistic food photography integrated into a clean grid.
- + Strong adherence to font style and sectioning requirements.
- − Nonsensical placeholder text for the dish names and descriptions.
- − Repetition of the 'Appetizers' header instead of including 'Pizza' as a separate section header.
Stable Diffusion 3.5 Large Turbo
- + Creative interpretation with high-contrast colors.
- + Good legibility of section headers like 'Pizza' and 'Mians'.
- + Unique grid-style presentation of food dishes.
- − Poor composition that feels like a collection of objects rather than a menu.
- − High degree of artifacts in the food illustrations, making them look plastic or AI-distorted.
- − Spelling errors in main headers ('Mians' instead of 'Mains').
Verdict: FLUX.1 Krea [dev] is the clear winner as it produces a professional, cohesive menu layout that matches the 'modern minimalist' prompt perfectly. Stable Diffusion 3.5 Large Turbo fails to create a unified design, instead scattering menu cards and distorted food icons across a white space, resulting in a cluttered and unrealistic aesthetic.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent text rendering with accurate spelling for all requested phrases.
- + Follows the exploding burger concept with clear separation between components.
- + Professional photorealistic lighting and realistic food textures.
- − The price '€6.99' is not inside a starburst as requested.
- − The glowing/fiery effect on the text is relatively subtle compared to the prompt's request.
Stable Diffusion 3.5 Large Turbo
- + Strong sense of motion and fiery atmosphere with glowing embers.
- + Creatively incorporates the fire into the base of the burger.
- − Completely failed to include any of the requested text elements.
- − The burger components are not truly 'exploded' or suspended separately; it looks like a stacked burger with a stick through it.
- − Visual quality is less photorealistic and more like a digital illustration or render.
Verdict: FLUX.1 Krea [dev] is the clear winner as it successfully integrated all the complex text requirements with perfect spelling, whereas Stable Diffusion 3.5 Large Turbo failed to include any text at all. FLUX.1 Krea [dev] also adhered much better to the 'exploded' composition, providing a professional-grade advertisement look compared to the more illustrative style of Stable Diffusion 3.5 Large Turbo.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent text legibility and accuracy
- + Realistic chalk texture and board smears
- + Authentic cursive and print handwriting styles
- − Included extra menu items not in the prompt
- − Text rendering became slightly jumbled in the middle lines ('Gruffle', 'Coman')
Stable Diffusion 3.5 Large Turbo
- + Aesthetically pleasing café environment
- + Good lighting and composition
- − Severe spelling errors and non-existent words
- − Failed to follow the date and price instructions correctly
- − The font looks like a digital brush rather than realistic chalk
Verdict: FLUX.1 Krea [dev] significantly outperformed Stable Diffusion 3.5 Large Turbo in text rendering and prompt adherence, successfully capturing the specific date and most menu items with a realistic chalk texture. While Model A had some minor spelling hiccups, Model B produced largely illegible 'gibberish' text and failed to follow the requested layout and content.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.1 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent photographic realism and cinematic lighting
- + Creative interpretation of the prompt with the astronaut integrated into the horse's gear
- + High level of detail in the space suit and Earth's surface
- − Failed the primary logic constraint; the human is on top of the horse
Stable Diffusion 3.5 Large Turbo
- + Strong composition with a sense of motion
- + Good contrast and clean lines
- + High resolution with smooth textures
- − Failed the primary logic constraint; the human is on top of the horse
- − The small planet in the upper left has unnatural smearing/interpolation artifacts
Verdict: Both models failed the specific 'horse on top, not vice versa' instruction, which was a logic test for the prompt. FLUX.1 Krea (dev) is the winner because its visual quality, cinematic lighting, and detailed textures are significantly more realistic and impressive than the flatter, more artificial look of Stable Diffusion 3.5 Large Turbo.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent adherence to the scene description, including the passenger with a phone.
- + The capybara's expression is perfectly calm and professional as requested.
- + Realistic textures on the capybara's fur and the leather steering wheel.
- − Includes an extra person in the front passenger seat which was not requested.
- − The taxicab sign on the roof appears to be inside the door frame or awkwardly positioned.
Stable Diffusion 3.5 Large Turbo
- + Dynamic lighting with vibrant bokeh in the background.
- + Correctly depicts the capybara's front paws on the steering wheel.
- + The capybara's hat and jacket are well-rendered and fit the theme.
- − Completely missed the passenger in the back seat looking at her phone.
- − The text on the cap's logo is garbled and unreadable.
- − The capybara's anatomy around the neck and shoulder area looks somewhat distorted.
Verdict: FLUX.1 Krea (dev) followed the complex multi-character prompt much more effectively, even though it added an extra person in the front; it successfully captured the 'bored businesswoman on her phone' detail. Stable Diffusion 3.5 Large Turbo produced a visually striking image of the capybara but ignored the passenger entirely, failing a primary component of the requested scene.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent gothic atmosphere with cinematic lighting and twisted silhouettes.
- + Includes all requested text elements, including a complex scroll banner.
- + The border design with webs and thorns is intricate and fits the vintage theme well.
- − Several spelling errors and typos in the generated text (e.g., 'Pasty', 'nights tright', 'are are').
- − The composition feels slightly cluttered around the text areas.
Stable Diffusion 3.5 Large Turbo
- + Clean, readable text for the main title with no major spelling errors.
- + The central jack-o-lantern is well-defined and has a strong glow effect.
- − Completely failed to include the scroll banner and the specific event details (Date, Time, Location).
- − The visual style is more cartoonish/vector-like rather than the requested 'vintage gothic' or 'cinematic' look.
- − The 'moody night sky' is replaced by a bright, flat yellow background.
Verdict: FLUX.1 Krea [dev] followed the prompt much more closely, including the background atmosphere, scroll banners, and specific event details, although it suffered from several spelling hallucinations. Stable Diffusion 3.5 Large Turbo failed to include almost all the secondary text requirements and the parchment aesthetic, resulting in a generic card that lacks the requested gothic depth.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent typography with perfect spelling and placement.
- + Superb clean, professional lighting and shadows.
- + Followed the background color request perfectly.
- − The flag icon is a bit disconnected from the diorama base concept.
Stable Diffusion 3.5 Large Turbo
- + Nice interpretation of the 'diorama base' as a wooden block.
- + Good 3D cartoon textures for the sushi components.
- − Text spelling error on 'SUSHI' (rendered as 'SIIHI').
- − Failed to provide a solid light blue background.
- − The flag icon is incorrect/unrecognizable.
Verdict: FLUX.1 Krea (dev) followed every instruction in the prompt perfectly, delivering high-quality typography, the correct background color, and a clean minimalist aesthetic. Stable Diffusion 3.5 Large Turbo struggled with the text rendering, failed the background color requirement, and produced an incorrect flag.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Followed the prompt completely by including all four distinct animals: puppy, kitten, bunny, and fox.
- + Excellent composition with a sense of motion and playfulness in a wide meadow.
- + Very high visual quality with realistic fur textures and beautiful backlighting/god rays.
- − Includes two kittens instead of one kitten and one bunny.
- − The kitten on the right appears to be floating somewhat unnaturally.
Stable Diffusion 3.5 Large Turbo
- + High contrast with vibrant colors and clear expressive eyes.
- + Backlighting on the fur is well-defined and aesthetically pleasing.
- − Failed to include all requested animals, missing the bunny and the fox.
- − The animals look more like stylized illustrations or 3D renders than hyper-photorealistic.
- − The dog features appear distorted, particularly the fur on the ears and the neck blending.
Verdict: FLUX.1 Krea [dev] is the clear winner as it attempted all four animal types and successfully captured the feeling of a wide, sunlit meadow with realistic textures. In contrast, Stable Diffusion 3.5 Large Turbo failed to include half of the requested subjects and produced a much more artificial, digital-looking image that lacked the 'masterpiece' quality requested.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Excellent hand-drawn etching style that fits the vintage aesthetic.
- + Beautifully balanced composition with a classic banner layout.
- + Great use of subtle paper texture and distressed edges.
- − Misspelled the primary text as 'Café Flanelín'.
- − Includes some illegible decorative text at the bottom.
Stable Diffusion 3.5 Large Turbo
- + Successfully spelled 'Caffeé Florin' (though with an extra 'e').
- + Solid vector-like shapes with a clean, high-contrast look.
- + Good adherence to the color palette requested.
- − Visual artifacting where the 'n' in Florian merges into the ribbon.
- − The cloche dome is stylized into a strange cocktail shaker or mug hybrid.
- − The steam lines are somewhat generic and less artistic than Model A.
Verdict: FLUX.1 Krea (dev) captures the 'vintage minimalist' and 'vector emblem' style much more effectively with its beautiful woodcut-style illustration, but it fails significantly on the spelling of the requested name. Stable Diffusion 3.5 Large Turbo adheres better to the text requirements but suffers from odd object merging and a less sophisticated artistic execution. FLUX.1 Krea is the preferred choice for a professional brand aesthetic despite the typo.
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 Krea [dev]
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.1 Krea [dev]
- + Successfully rendered 'Apollo 11' text clearly
- + Includes a sequence of numbered steps as requested
- + Uses the requested NASA-inspired navy and white palette
- − Failed to include a Saturn V rocket or Lunar Module icon
- − Icons and numbering are nonsensical (e.g., number 4 is four humans, number 1 is Saturn)
- − Layout is disorganized and lacks professional infographic flow
Stable Diffusion 3.5 Large Turbo
- + Strong professional infographic layout with sidebar and main visual
- + Captures the flat-vector aesthetic and 'NASA' vintage feeling perfectly
- + Includes specific trajectory lines and moon surface details
- − Failed to include all 6 specific steps requested
- − Text rendering is poor with several spelling errors ('Apoll.o', 'Laurch')
- − Inaccurate icons for the sidebar steps
Verdict: Both models struggled with following the specific 6-step sequences and iconography requested in the prompt. FLUX.1 Krea produced a more legible title and basic numbered list but failed on the visual style and icon accuracy, while Stable Diffusion 3.5 Large Turbo created a much more visually compelling and stylistically accurate 'NASA' poster that suffered from poor text spelling and missing steps.
Next steps
Explore each model
Distilled version of SD 3.5 Large that generates high-quality images in just 4 steps, offering faster inference and reduced costs