Black Forest Labs' distilled 9 billion parameter image generation model with sub-second inference and multi-reference support
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.2 [klein] 9B
20.6
arena score
#13 of 32 in Image Editing
Skill signature
· Image Editing
Stable Diffusion 3.5 Large Turbo
10.7
arena score
#61 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.2 [klein] 9B
100.0%
win rate
Ties
0.0%
Stable Diffusion 3.5 Large Turbo
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.2 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent photorealism with natural textures on the book and table.
- + Perfect spatial adherence to the prompt with the blue sphere inside the cube and book on top.
- + Advanced light interactions, including caustics on the table and refraction through the glass.
- − The glass cube has rounded edges and thick walls, making it look more like a heavy vase or container.
Stable Diffusion 3.5 Large Turbo
- + Clean, modern aesthetic with sharp lines.
- + Good lighting and shadow work on the table surface.
- − Failed the spatial aspect of the prompt; the book is inside the cube rather than on top of it.
- − The blue sphere is sitting on the book rather than primarily inside the cube as an independent object.
- − The plant is to the left rather than behind the cube, failing to show through the glass.
Verdict: FLUX.2 [klein] 9B is the clear winner as it followed every spatial instruction in the prompt perfectly, including complex details like the plant being visible through the glass. In contrast, Stable Diffusion 3.5 Large Turbo failed to place the book on top of the cube and placed the plant in the wrong position, although it produced a clean digital render.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Exceptional realism with highly detailed skin textures and fabric.
- + Accurately represents the 50mm shallow depth of field and bokeh.
- + Captures the atmosphere of light rain and street reflections perfectly.
- − The hands show some minor anatomical blurring/merging.
- − The bicycle chain and pedal geometry is slightly physically impossible.
Stable Diffusion 3.5 Large Turbo
- + Includes a red bicycle as requested.
- + Attempts to show rain via vertical streaks.
- − Anatomical failure with the subject's head appearing detached or incorrectly positioned.
- − Flat, plastic-like textures that lack the requested natural skin realism.
- − Completely ignores the 'motion blur from passing cars' and 'shallow depth of field' requirements.
Verdict: FLUX.2 [klein] 9B followed the prompt with high fidelity, creating a cinematic and believable scene with excellent texture work and atmospheric effects. In contrast, Stable Diffusion 3.5 Large Turbo failed significantly on anatomy and realism, producing a flat, low-detail image that missed key technical camera instructions.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent adherence to all prompt details including beads in hair and leather textures.
- + Superior photographic realism with believable skin texture and lighting.
- + High-quality engraving details on the plate armor that look physically embossed.
- − The sparks look slightly uniform and a bit like graphic overlays.
Stable Diffusion 3.5 Large Turbo
- + Strong composition with a dynamic angle and good lighting on the face.
- + Captures the 'battle-worn' aspect with clear blood/dirt stains.
- − Failed to include 'small beads' in the hair as requested.
- − The skin and hair have a smooth, almost 'plastic' CGI appearance compared to the realism of Model A.
- − Lighting feels disjointed, with very harsh white glints that don't match the warm torchlight prompt.
Verdict: FLUX.2 [klein] 9B followed the prompt significantly better than Stable Diffusion 3.5 Large Turbo, accurately incorporating the specific details about hair beads and complex skin textures. While Stable Diffusion 3.5 Large Turbo produced a striking character portrait, it suffered from an artificial, smooth texture and failed to render several of the specific material requirements.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Strong typography that resembles a real menu layout
- + Accurate category headers for 'Appetizers', 'Pizza', and 'Mains'
- + Clean, professional grid composition that follows graphic design principles
- − Nonsense 'lorem ipsum' style text is messy and glitched in places
- − Some redundant headers like double 'Appetizers' and 'Mains'
Stable Diffusion 3.5 Large Turbo
- + Vibrant food photography with good color contrast
- + Clean white background with a modern aesthetic
- + Successful grid layout for the food imagery
- − Typos in section headers such as 'Mians'
- − The food items look slightly artificial and plastic-like
- − The separate menu cards below the images feel disconnected from the main design
Verdict: FLUX.2 [klein] 9B produces a much more convincing graphic design layout that actually looks like a printed menu, despite some nonsensical text strings. Stable Diffusion 3.5 Large Turbo provides better food photography, but the overall layout is disjointed and contains more obvious spelling errors in the headers.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent typography rendering with the requested fiery effect and perfect spelling.
- + Highly realistic food textures on the bun, patties, and vegetables.
- + Successfully followed all prompt instructions including the price starburst and exploded composition.
- − The 'exploded' effect is slightly conservative, as components are still largely stacked together.
Stable Diffusion 3.5 Large Turbo
- + Strong sense of heat and fire with effective use of smoke and glowing embers.
- + Dynamic lighting that creates a high-contrast, dramatic atmosphere.
- − Completely failed to include the requested text ('MAGIC BURGER', etc.).
- − The burger anatomy is messy with unrealistic dripping black spheres and a wooden skewer not requested.
- − The burger is not truly 'exploded' into individual suspended components.
Verdict: FLUX.2 [klein] 9B is the clear winner as it followed every instruction in the prompt, including complex text integration and specific price formatting. In contrast, Stable Diffusion 3.5 Large Turbo completely ignored the text requirements and produced a less organized image with several visual artifacts.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent text rendering with perfect spelling and realistic chalk texture.
- + Authentic hand-drawn aesthetic with smudge marks and natural variations.
- + Strict adherence to the specific menu items and date requested in the prompt.
- − Minor typo in the bottom fine print where 'fresh' is spelled 'fress'.
- − The composition is a tight crop on the board rather than showing the café environment.
Stable Diffusion 3.5 Large Turbo
- + Provides a broader environmental context showing the café interior and lighting.
- + The composition is balanced and clean.
- − Significant spelling errors and gibberish text throughout the menu.
- − The lettering looks like a smooth digital font rather than realistic chalk texture.
- − Fails to follow the specific date and menu items requested in the prompt.
Verdict: FLUX.2 [klein] 9B followed the prompt instructions meticulously, delivering highly accurate text with a convincing chalk-on-blackboard texture. In contrast, Stable Diffusion 3.5 Large Turbo failed on almost every textual requirement, producing numerous typos and utilizing a digital-looking font that lacked the requested hand-written authenticity.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.2 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent level of detail in the cosmic environment and textures
- + Highly cinematic lighting and composition
- + Successful rendering of realistic facial features through the visor
- − Failed to follow the surreal component of the prompt ('horse on top, not vice versa')
- − Includes prominent nonsensical AI-generated text/watermarking
Stable Diffusion 3.5 Large Turbo
- + Dynamic sense of scale and motion
- + Clean rendering of the white space suit and horse
- + Good contrast between the subjects and the dark space background
- − Failed to follow the specific spatial instruction ('horse on top, not vice versa')
- − The anatomy of the horse's front legs is distorted
- − The 'Earth' in the top-left appears low resolution and blurry
Verdict: Both models failed the specific logic test requested in the prompt ('horse on top, not vice versa'), instead providing the standard astronaut-on-horse image. FLUX.2 [klein] 9B is the preferred image because it delivers a much higher level of detail, a more cinematic atmosphere, and a better rendering of the cosmic scene, despite the inclusion of unwanted text.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent photorealism in the capybara's fur and the interior lighting.
- + Perfect adherence to the businesswoman's bored expression and her position in the back seat.
- + Accurate rendering of the New York taxi driver cap and jacket.
- − The perspective looks more like a modern compact car than a standard Crown Victoria taxi interior.
- − Minor text artifacts on the hat ('TALA').
Stable Diffusion 3.5 Large Turbo
- + Dynamic lighting with attractive bokeh in the background.
- + Correct placement of front paws on the steering wheel.
- + The capybara's expression is calm and professional as requested.
- − The businesswoman in the back is blurry and lacks details.
- − The passenger is not visibly looking at a phone.
- − The capybara's fur has a slightly more illustrative/digital art feel compared to Image A.
Verdict: FLUX.2 [klein] 9B is the clear winner as it followed every instruction in the prompt, including the specific boredom of the passenger and her engagement with a phone. While Stable Diffusion 3.5 Large Turbo created a visually striking image with nice bokeh, it failed to clearly depict the businesswoman and her activities in the back seat.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Perfect adherence to all text requirements including title, scroll banner, and event details.
- + Excellent composition with a rich, cinematic atmosphere and high-quality vintage textures.
- + Strong gothic aesthetic that perfectly matches the 'dark parchment' and 'thorns' descriptors.
- − The placement of 'Invitation' in the title is slightly cramped against the jack-o-lantern stem.
Stable Diffusion 3.5 Large Turbo
- + Consistent jack-o-lantern designs with good glowing effects.
- − Failed to include the required scroll banner and event specific details (date, time, location).
- − Missing 'Invitation' from the main title and features a typo in 'Party'.
- − Flat, clip-art style composition that lacks the requested cinematic and vintage parchment feel.
Verdict: FLUX.2 [klein] 9B followed every instruction perfectly, producing a professional-grade invitation with accurate text and a dark, moody atmosphere. In contrast, Stable Diffusion 3.5 Large Turbo failed to include the event details and banner, while also suffering from spelling errors and a much simpler, less atmospheric visual style.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent typography with perfect spelling and alignment.
- + Clean, professional composition that adheres closely to the 'minimal' request.
- + Soft, pleasing lighting and high-quality textures on the wooden base and sushi.
- − The flag icon is incorrect, resembling the flag of Yemen rather than Japan.
- − The sushi modeling is slightly simplistic compared to the realistic PBR request.
Stable Diffusion 3.5 Large Turbo
- + Strong 3D isometric aesthetic with good depth and shadow work.
- + Vibrant colors and detailed texture on the rice grains.
- − Incorrect spelling of 'SUSHI' as 'SIIHI'.
- − The text and flag are rendered as 3D physical props instead of the requested top-center floating text.
- − The flag icon is logically incorrect for Japan.
Verdict: FLUX.2 [klein] 9B is the clear winner due to its superior text rendering and adherence to the layout instructions, placing the text at the top-center as requested rather than as physical 3D objects. While both models failed to generate a correct Japanese flag, Stable Diffusion 3.5 Large Turbo suffered from significant spelling errors and ignored the 'minimal' and 'solid light blue background' constraints.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent adherence to the wildlife meadow setting with realistic flowers and god rays.
- + Captures three out of the four animals requested with high detail and character.
- + Beautiful lighting and sparkling bokeh effects that align with the sunrise theme.
- − Missed the baby bunny requested in the prompt.
- − The kitten's pose and anatomy around the front paws are slightly unnatural.
Stable Diffusion 3.5 Large Turbo
- + Strong 'big expressive eyes' and soft fur textures.
- + Good use of backlight to highlight fur silhouettes.
- − Failed to include the fox and the bunny, providing only cat/dog hybrids.
- − Significant anatomical issues, notably the floating/extra paw on the dog's chest.
- − The environment is generic and lacks the 'lush wildflower meadow' detail of the other image.
Verdict: FLUX.2 [klein] 9B followed the prompt more effectively, capturing the meadow environment and three of the four specified animals with high clarity. Stable Diffusion 3.5 Large Turbo struggled with the complex list of animals, resulting in anatomical errors and a much simpler composition that missed several key prompt elements.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Excellent typography with correct Italian accent
- + Clean vector aesthetic that perfectly matches the minimalist requirement
- + Strong brand balance and professional logo composition
- − The steam icon is slightly simplistic compared to the rest of the illustration
Stable Diffusion 3.5 Large Turbo
- + Beautifully detailed shading and highlights
- + Classic vintage engraving style
- + Good use of textured background
- − Spelling error in 'Caffeé' (extra 'e' and wrong accent direction)
- − Text artifacting/jitter on the letter 'n'
- − Less minimalist than requested, leaning more into complex illustration
Verdict: FLUX.2 [klein] 9B followed the prompt perfectly, delivering a clean, professional vector logo with correct spelling and a balanced composition. Stable Diffusion 3.5 Large Turbo produced a more illustrative and visually striking image, but failed on the basic requirement of text accuracy and exceeded the 'minimalist' constraint.
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 [klein] 9B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 9B
- + Strong adherence to the flat-vector style with crisp lines.
- + Includes all requested steps from launch to landing with specific icons.
- + Better legibility of main headings compared to the competitor.
- − Several spelling errors in the labels (e.g., 'EARDHT', 'LANDINING', 'ARMSTROMG').
- − Flow of steps is somewhat disjointed and lacks a clear sequence path.
Stable Diffusion 3.5 Large Turbo
- + Sophisticated composition and layout that feels like a professional poster.
- + Excellent adherence to the requested NASA-inspired color palette.
- + Includes creative extra details like the astronaut and planetary horizons.
- − Fails to include all six specific infographic steps requested.
- − Text is largely illegible/gibberish beyond the titles.
- − The iconography is abstract and does not clearly match the Saturn V or Lunar Module requests.
Verdict: FLUX.2 [klein] 9B followed the prompt's structural requirements more closely by including the specific steps of the mission and distinct icons, despite having several typos. Stable Diffusion 3.5 Large Turbo produced a more aesthetically pleasing and balanced poster with a superior color palette, but it failed significantly on the infographic's content density and specific step-by-step requirements.
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