Black Forest Labs' flagship image generation model delivering state-of-the-art quality with exceptional realism, precision, and consistency for both text-to-image and advanced image editing
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.2 [max]
25.9
arena score
#10 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Stable Diffusion 3.5 Large
22.8
arena score
#30 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.2 [max]
77.3%
win rate
Ties
0.0%
Stable Diffusion 3.5 Large
22.7%
win rate
77.3%
0.0% ties
22.7%
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.2 [max]
Stable Diffusion 3.5 Large
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [max]
- + Perfect adherence to the spatial arrangement requested, with the book sitting on top of the cube.
- + Excellent realism in textures, particularly the leather-bound book and the glass refractive properties.
- + Soft lighting correctly interacts with the glass, creating realistic internal reflections and shadows on the table.
- − The plant in the background is quite blurred, making it slightly less distinct as being 'behind' the glass.
Stable Diffusion 3.5 Large
- + Accurate rendering of the requested elements including the blue sphere and wooden table.
- + The plant is clearly visible through the glass as requested.
- − Failed the spatial instruction 'On top of the cube sits a red book'; instead, it placed the cube on top of the book.
- − The lighting is somewhat harsh and inconsistent with 'soft window light'.
- − Visible artifacts on the edges of the glass cube where it meets the book.
Verdict: FLUX.2 [max] followed the complex spatial instructions perfectly, placing the red book on top of the glass cube and the sphere inside. Stable Diffusion 3.5 Large reversed the order of the cube and the book, which also resulted in the sphere appearing to float or sit on the book rather than being inside the cube.
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.2 [max]
Stable Diffusion 3.5 Large
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [max]
- + Excellent skin texture and realistic age details on the hands and face.
- + Perfect adherence to the shallow depth of field and motion blur requirements.
- + Highly realistic bicycle anatomy and wet pavement reflections.
- − The 'imperfect framing' request is subtle, as the composition feels quite professionally balanced.
Stable Diffusion 3.5 Large
- + Includes a bus and car in the background to establish the street scene.
- + Good color contrast with the red bicycle against the cool tones.
- − The man's hands are fused into a singular mass of flesh, lacking fingers.
- − The bicycle frame geometry is illogical and broken near the pedals.
- − The rain effect looks like a digital overlay rather than an environmental element.
Verdict: FLUX.2 [max] produced a nearly photorealistic image that perfectly captured the technical requirements like 50mm lens feel, motion blur, and natural skin textures. Stable Diffusion 3.5 Large struggled significantly with anatomical correctness, resulting in deformed hands and a structurally unsound bicycle, while also failing to match the level of photographic realism requested.
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.2 [max]
Stable Diffusion 3.5 Large
AI Judge Analysis
FLUX.2 [max]
- + Excellent adherence to the 'hair braided with small beads' requirement.
- + Highly realistic skin textures with convincing scars and dirt.
- + Superior cinematic lighting and depth of field execution.
- − The leather strap across the chest has slightly muddy texture compared to the metalwork.
Stable Diffusion 3.5 Large
- + Intricate engraving detail on the plate armor.
- + Strong character expression and intense gaze.
- + Good interpretation of the cloth underlayer with the visible chainmail/quilted texture.
- − Failed to include the requested beads in the braided hair.
- − Lighting feels more like daylight than the requested warm torchlight.
Verdict: FLUX.2 [max] followed the prompt more closely by accurately including the beads in the hair and creating a distinct warm torchlight atmosphere. While Stable Diffusion 3.5 Large produced beautiful armor engravings, it missed specific character details and the requested lighting mood, favoring a brighter outdoor-style scene.
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.2 [max]
Stable Diffusion 3.5 Large
50% wins
0% ties
50% wins
AI Judge Analysis
FLUX.2 [max]
- + Excellent adherence to all required sections (Appetizers/Pizza/Mains).
- + Clean, professional layout that genuinely looks like a usable restaurant menu.
- + Higher text legibility and better organization of price points.
- − Small spelling artifacts in item descriptions.
- − The food photos on the left are a vertical column rather than a full grid across the page.
Stable Diffusion 3.5 Large
- + High-quality, vibrant food photography in a clear grid layout.
- + Bold, modern sans-serif typography for the main headers.
- − Poor text rendering for smaller details, resulting in illegible 'garbled' characters.
- − Layout is less practical for a menu, with sections compressed into a narrow center column.
- − Misspelled key headers (e.g., 'MAIMAES' instead of Mains, 'APPETIZRS').
Verdict: FLUX.2 [max] is the superior choice because it produces a functional, logical menu layout that correctly incorporates all requested sections with professional spacing. In contrast, Stable Diffusion 3.5 Large creates a visually striking grid but fails significantly on text legibility and logical structure, making the menu unusable.
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.2 [max]
Stable Diffusion 3.5 Large
AI Judge Analysis
FLUX.2 [max]
- + Perfectly follows the exploded view requirement with all components suspended.
- + Excellent text integration with accurate spelling, font effects, and starburst placement.
- + Professional commercial lighting and high-quality photorealistic textures on the ingredients.
- − The 'fiery' theme in the background is a bit subtle compared to the burger lighting.
- − Some sauce droplets look slightly static rather than having a high sense of motion.
Stable Diffusion 3.5 Large
- + Dramatic and vibrant fiery background with actual burning coals and intense light.
- + High level of appetizing detail on the dripping cheese and charred texture of the patties.
- − Completely failed to include any of the requested text elements.
- − Failed the 'exploded' instruction as the burger is mostly assembled rather than suspended in mid-air.
- − The top bun shows some anatomical warping where the sesame seeds meet the edge.
Verdict: FLUX.2 [max] is the clear winner as it successfully integrated all complex prompt requirements, including several specific text strings and a difficult 'exploded' layout. Stable Diffusion 3.5 Large produced a visually striking image with great fire effects, but it ignored the text commands entirely and failed to separate the burger components as requested.
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.2 [max]
Stable Diffusion 3.5 Large
AI Judge Analysis
FLUX.2 [max]
- + Excellent text rendering with no spelling errors
- + Authentic chalk texture and natural handwriting variations
- + Strict adherence to the requested date and menu prices
- − The 'Brown Butter' item is truncated slightly early based on the prompt's cut-off, though it completes the logical thought
Stable Diffusion 3.5 Large
- + Provides a wider environmental context of a café interior
- + Good composition and lighting in the room
- − Numerous spelling mistakes such as 'TODAAY' and 'Cholcalte'
- − Failed to follow the date (2024 instead of 2026)
- − Text layout is cluttered and looks like a digital font rather than natural handwriting
Verdict: FLUX.2 [max] significantly outperformed the competition by rendering the requested text with near-perfect accuracy and a realistic chalk texture. Stable Diffusion 3.5 Large struggled with the text requirements, producing multiple misspellings and failing to follow the specific date and layout instructions from the prompt.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.2 [max]
Stable Diffusion 3.5 Large
AI Judge Analysis
FLUX.2 [max]
- + Excellent anatomical detail on the horse and space suit.
- + Great lighting and composition with the background galaxy.
- + High clarity and resolution in textures.
- − Completely failed the negative constraint; the astronaut is riding the horse.
Stable Diffusion 3.5 Large
- + Atmospheric and cinematic lighting effects.
- + Dynamic sense of motion with the cosmic dust.
- − Completely failed the negative constraint; the astronaut is riding the horse.
- − The horse's front legs have anatomical issues and strange blurring.
Verdict: Both FLUX.2 [max] and Stable Diffusion 3.5 Large failed to follow the specific spatial instruction 'horse on top, not vice versa,' instead producing the cliché 'astronaut riding a horse' image. Between the two, FLUX.2 [max] is the superior image due to much cleaner rendering, better horse anatomy, and higher overall detail compared to the messy artifacts seen in Stable Diffusion 3.5 Large's horse legs.
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.2 [max]
Stable Diffusion 3.5 Large
AI Judge Analysis
FLUX.2 [max]
- + Excellent adherence to the complex multi-character prompt by including the businesswoman in the back seat.
- + Highly realistic lighting and cinematic composition inside the taxi.
- + Natural integration of the capybara within the car's interior.
- − The capybara has distinctly human hands on the steering wheel, which is a significant anatomical error.
- − The capybara's head is slightly disconnected from the human-like neck/body structure.
Stable Diffusion 3.5 Large
- + Correctly renders capybara paws rather than human hands.
- + The animal's fur and whiskers have high texture quality.
- + Vibrant colors and sharp details on the taxi door and jacket.
- − Fails to include the requested human businesswoman in the back seat.
- − The capybara is not touching the steering wheel, despite the prompt's instruction.
- − The scene feels less like New York at night and more like a studio shot.
Verdict: FLUX.2 [max] successfully composed the entire scene, including the businesswoman and her expression, which Stable Diffusion 3.5 Large completely omitted. While FLUX.2 [max] suffers from anatomical issues by giving the capybara human hands, it is the superior choice for overall prompt adherence and scene-building, whereas Stable Diffusion 3.5 Large failed on several core instructions.
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.2 [max]
Stable Diffusion 3.5 Large
AI Judge Analysis
FLUX.2 [max]
- + Perfect text rendering for all requested details including the date, time, and location.
- + Excellent cinematic lighting and high-fidelity textures on the pumpkin and parchment.
- + Balanced composition that remains elegant and readable despite the gothic elements.
- − The 'dark parchment' feel is limited to the outer border rather than the whole background.
- − The scroll banner is somewhat small relative to the other text elements.
Stable Diffusion 3.5 Large
- + Strong interpretation of the 'dark parchment poster' aesthetic with torn edges.
- + Good inclusion of twisted trees and a moody, glowing moon for atmosphere.
- − Completely failed to include the event details (Date, Time, Location) at the bottom.
- − Text rendering on the scroll banner is messy with artifacts and incomplete words.
- − The jack-o-lantern is not the central focus as requested, appearing off to the side instead.
Verdict: FLUX.2 [max] is the clear winner as it successfully rendered all requested text, including the specific date and location, with perfect clarity. Stable Diffusion 3.5 Large missed the bottom half of the prompt's instructions and struggled with text legibility on the scroll banner.
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.2 [max]
Stable Diffusion 3.5 Large
50% wins
0% ties
50% wins
AI Judge Analysis
FLUX.2 [max]
- + Perfectly followed layout instructions with text at the top-center and a solid light blue background.
- + The 45-degree isometric perspective is precise and clean.
- + Exceptional minimalist 3D rendering with soft, refined textures and realistic lighting.
- − The sushi variety is a bit simple compared to the other model.
Stable Diffusion 3.5 Large
- + High level of detail in the sushi models, specifically the rice grain textures.
- + Vibrant and appealing color palette.
- − Failed to place text at the top-center, instead attaching it to a flag within the scene.
- − Ignored the 'minimal garnish' instruction, creating a cluttered compositions with many decorative elements.
- − Background has a slight gradient/shadow rather than being a solid color.
Verdict: FLUX.2 [max] followed every aspect of the prompt, including the specific text placement, isometric angle, and minimalist aesthetic. In contrast, Stable Diffusion 3.5 Large produced a much more cluttered scene that integrated the text into the objects rather than placing it as an overlay, and it largely ignored the request for a 'minimal' diorama.
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.2 [max]
Stable Diffusion 3.5 Large
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [max]
- + Perfect adherence to all four requested animal species (golden retriever, tabby kitten, bunny, red fox).
- + Ultra-detailed fur textures and anatomy, particularly on the fox's tail and the kitten's paws.
- + Balanced composition with clear separation between the subjects and the environment.
- − The lighting, while warm, lacks the explicit 'god rays' effect requested in the prompt.
- − The kitten is significantly smaller than the bunny, which looks slightly out of scale.
Stable Diffusion 3.5 Large
- + Excellent capture of the 'joyful' vibe with more expressive, smiling faces on the puppy and fox.
- + Stronger interpretation of the golden sunrise and 'god rays' lighting effects.
- + Good sense of motion and action with the animals running toward the camera.
- − Incomplete prompt adherence; it missed the 'tabby kitten' and replaced it with a generic ginger feline.
- − Noticeable anatomical artifacts, such as the fox kit's lack of distinctive black paws and the puppy's front left leg looking slightly mangled.
- − Heavy bokeh/blur makes some of the butterflies and background elements look messy.
Verdict: FLUX.2 [max] is the winner due to its superior anatomical accuracy and strict adherence to the requested animal list, including the specific tabby markings and fox features. While Stable Diffusion 3.5 Large captured a more energetic and well-lit 'joyful' atmosphere, it failed to render a tabby kitten and suffered from blurred details and slight anatomical distortions.
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.2 [max]
Stable Diffusion 3.5 Large
80% wins
0% ties
20% wins
AI Judge Analysis
FLUX.2 [max]
- + Perfect text rendering for both name and date banner
- + Clean, balanced vector emblem composition
- + Excellent adherence to the vintage minimalist aesthetic
- − The steam lines are very thin compared to the rest of the illustration
Stable Diffusion 3.5 Large
- + Nice use of negative space in the cloche design
- + Adds decorative corner flourishes to the background
- − Misspelled the name as 'Cafféé Florian' with an extra 'e'
- − The cloche is disconnected and floating awkwardly
- − The steam coming from the top of the dome is illogical
Verdict: FLUX.2 [max] significantly outperformed Stable Diffusion 3.5 Large by correctly spelling 'Caffè Florian' and creating a cohesive, professional vector emblem. Stable Diffusion 3.5 Large suffered from typographical errors and a disjointed illustration where the cloche lid floats above the base with no clear connection.
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.2 [max]
Stable Diffusion 3.5 Large
50% wins
0% ties
50% wins
AI Judge Analysis
FLUX.2 [max]
- + Excellent adherence to the logical flow of the requested 6 steps.
- + Clean, legible typography with accurate spelling for the most part (except one small typo).
- + Consistent, high-quality flat-vector iconography that matches the NASA-inspired theme perfectly.
- − Small typo 'Tranquiity' instead of 'Tranquility'.
- − The layout of the steps is slightly non-linear (jumping from top row to bottom middle for Translunar).
Stable Diffusion 3.5 Large
- + Detailed vector-style illustrations with a nice vintage aesthetic.
- + Good use of the requested color palette.
- − Completely fails to follow the logical 6-step infographic structure requested.
- − Text is garbled and unreadable, featuring many 'gibberish' characters.
- − Inaccurate imagery, such as depicting a Space Shuttle-style vehicle instead of a Saturn V for Apollo 11.
Verdict: FLUX.2 [max] followed the prompt instructions precisely, creating a logical, readable, and aesthetically pleasing infographic with clear steps and icons. Stable Diffusion 3.5 Large produced a messy layout with illegible text and a Space Shuttle that is historically inaccurate for the Apollo 11 mission.
Next steps
Explore each model
Stability AI's 8.1-billion parameter Multimodal Diffusion Transformer (MMDiT) text-to-image model featuring improved image quality, typography, complex prompt understanding, and resource-efficiency