Black Forest Labs' compact, open-source image generation model with sub-second inference, optimized for production and near real-time applications with multi-reference support
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.2 [klein] 4B
22.3
arena score
#32 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.2 [klein] 4B
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Perfect adherence to spatial instructions including the book sitting on top and the plant visible through the glass.
- + Photorealistic textures on the wood table and the book cover.
- + Natural lighting and realistic depth of field.
- − The glass cube has some minor artifacts on the edges where they meet the book.
Stable Diffusion 3.5 Large Turbo
- + Sleek, modern aesthetic with high contrast lighting.
- + Very clean geometry and sharp lines.
- − Failed the spatial prompt; the red book is inside the cube rather than on top.
- − The plant is not visible through the glass as requested.
- − The sphere placement feels less natural.
Verdict: FLUX.2 [klein] 4B followed every instruction in the complex spatial prompt perfectly, creating a highly realistic scene with the plant visible through the glass and the book correctly positioned on top. Stable Diffusion 3.5 Large Turbo failed to arrange the objects according to the prompt, placing the book inside the cube and the plant entirely to the side.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent adherence to the 'candid street photo' aesthetic with realistic lighting and rain effects.
- + Very high detail in skin texture, clothing, and the weathered bicycle.
- + Accurate representation of the requested 50mm shallow depth of field and urban background.
- − Physical logic error with the bicycle chain hanging loosely below the crankset.
- − The man's left hand is merged poorly with the bicycle handlebar area.
Stable Diffusion 3.5 Large Turbo
- + Successfully includes the red bicycle and elderly Japanese man as requested.
- + Clean colors and decent composition for a general illustration.
- − Serious anatomical and structural errors, particularly the head/hat merging into the torso.
- − Fails on the 'no stylization' and 'realistic' requirements, appearing more like a 3D render.
- − Lack of visible rain, motion blur, or realistic reflections on the pavement.
Verdict: FLUX.2 [klein] 4B followed the prompt's stylistic instructions perfectly, creating a gritty, cinematic, and highly detailed photograph that feels authentic despite minor mechanical errors in the bicycle. Stable Diffusion 3.5 Large Turbo failed to capture the realistic photographic quality requested, producing a plastic-looking image with significant anatomical distortions and a lack of environmental effects like rain or motion blur.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent anatomical realism and skin texture detail.
- + Perfectly captures the lighting interaction between the torches and the metal armor.
- + Strict adherence to all prompt details including the braided hair with beads and authentic scarring.
- − The composition is a bit static and centered.
- − The leather strap on the shoulder looks slightly flat compared to the metalwork.
Stable Diffusion 3.5 Large Turbo
- + Strong cinematic composition and lighting angle.
- + Good rendering of the ornate engraving on the shoulder plate.
- − Skin texture is overly smooth and looks plastic or airbrushed.
- − Failed to include the requested beads in the braided hair.
- − Lighting does not convincingly originate from the environment.
Verdict: FLUX.2 [klein] 4B is the clear winner as it provides a highly realistic, photograph-like quality that captures the requested 'battle-worn' texture through subtle scarring and dirt. In contrast, Stable Diffusion 3.5 Large Turbo produces a more stylized, 'video game' aesthetic with artificial skin smoothing and missed several specific prompt details like the hair beads.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Strong composition that resembles a real single-page menu layout.
- + Food photography is realistic and high-quality.
- + The grid layout for photography is well-integrated with the text sections.
- − The text is largely gibberish or nonsensical characters.
- − Includes spelling errors in large headers like 'MDAINS' and 'FRAZZA'.
Stable Diffusion 3.5 Large Turbo
- + Successfully spelled 'Pizza' correctly and 'Mians' is close to intended.
- + Uses a clean minimalist aesthetic with high-contrast colors.
- + Distinct separation between imagery and text lists.
- − The food photos look highly artificial and overly saturated.
- − The layout feels more like a mood board or flat-lay photograph than a functional menu design.
- − Odd floating elements and inconsistent perspectives in the food items.
Verdict: FLUX.2 [klein] 4B produces a much more professional and realistic menu layout that captures the essence of a modern restaurant design, despite the poor text rendering. Stable Diffusion 3.5 Large Turbo creates a decorative, 'cartoonish' top-down view that lacks the structured typographic hierarchy requested for a functional menu.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent photorealistic texture on the burger bun and meat.
- + High-quality starburst element with perfect price formatting.
- + Strong adherence to the background request with glowing embers and a dark atmosphere.
- − The main title 'MAGIC BURGER' is garbled and unreadable.
- − The burger is slightly tilted but not 'exploded' with suspended components as requested.
Stable Diffusion 3.5 Large Turbo
- + Successfully creative interpretation of a fiery atmosphere with smoke and flames.
- + Includes more distinct layers in the burger assembly.
- − Completely failed to include any of the requested text.
- − Image looks more like a digital illustration than the requested 'photorealistic detail'.
- − Did not adhere to the 'exploded' layout where components are suspended in mid-air.
Verdict: FLUX.2 [klein] 4B followed the prompt much more successfully than Stable Diffusion 3.5 Large Turbo, providing the secondary text and price starburst accurately despite the jumbled main title. Stable Diffusion 3.5 Large Turbo failed to include any text at all and produced a stylized, less realistic image that ignored several core prompt instructions.
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] 4B
Stable Diffusion 3.5 Large Turbo
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent chalk texture with realistic smudges and dust on the board.
- + Strong adherence to the requested text, completing the 'Brown Butter' portion from the truncated prompt.
- + Genuine handwritten appearance with variations in letter size and slant.
- − Several spelling errors in the menu items (e.g., 'Truffel Musheram', 'Ootrpous').
- − One extra 'S' in the word 'SPECIALS'.
Stable Diffusion 3.5 Large Turbo
- + Includes a 'cozy café' environmental context as requested in the prompt.
- + Attractive composition with lighting and plants.
- − Text consists largely of illegible or nonsensical gibberish.
- − The 'chalk' style looks like a digital brush rather than real chalk.
- − Failed to render the specific requested date and many prices incorrectly.
Verdict: FLUX.2 [klein] 4B followed the text instructions much more accurately, successfully interpreting the truncated 'Brown But...' as Brown Butter and maintaining a very realistic chalk texture. While it struggled with spelling, Stable Diffusion 3.5 Large Turbo failed significantly on text legibility, producing mostly gibberish that lacks the requested handwritten elegance.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.2 [klein] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + High resolution with sharp textures on the space suit and horse hair.
- + Beautifully detailed background with a realistic galaxy and planet curve.
- + Correct horse anatomy and lighting consistency.
- − Completely failed the negative constraint to have the horse on top, providing a standard astronaut riding a horse.
Stable Diffusion 3.5 Large Turbo
- + Clean, cinematic composition with a striking white-on-black aesthetic.
- + Good sense of motion and scale relative to the planet.
- − Failed the prompt's specific spatial instruction to place the horse on top of the astronaut.
- − Noticeable anatomy issues with the horse's legs and hooves.
Verdict: Both models completely failed to follow the specific spatial instruction 'horse on top, not vice versa,' opting instead for the stereotypical image of an astronaut riding a horse. FLUX.2 [klein] 4B is the winner based on purely visual grounds, as it offers superior detail, more realistic textures, and significantly better anatomy compared to the warped limbs in the Stable Diffusion 3.5 Large Turbo output.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent adherence to the 'businesswoman looking at phone' prompt requirement
- + Realistic taxi interior lighting and textures
- + The capybara's paws are correctly positioned on the steering wheel
- − The scale of the capybara is slightly too large compared to the car's interior
- − Background bokeh is a bit generic
Stable Diffusion 3.5 Large Turbo
- + Strong cinematic lighting with vibrant city bokeh
- + Highly detailed capybara fur and whiskers
- + The capybara's paws are anatomically interesting with visible digits
- − Failed to include the businesswoman in the back seat looking at her phone
- − Interior layout feels more like a modern car than a classic NYC taxi
Verdict: FLUX.2 [klein] 4B is the winner because it followed all instructions, specifically including the bored businesswoman in the backseat which is crucial for the surreal humor of the prompt. While Stable Diffusion 3.5 Large Turbo produced a very sharp image of the capybara, it missed the passenger and phone use entirely.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent cinematic lighting and spooky atmosphere.
- + Strictly follows the layout with the scroll banner and detailed text fields.
- + Includes all requested elements like thorns, webs, and twisted trees in a cohesive style.
- − Significant spelling errors in the main title ('Hallbwom Party niisation').
- − Minor typos in the event details ('30.10.206', 'Loecton').
Stable Diffusion 3.5 Large Turbo
- + Clean graphic design style with a very sharp central jack-o-lantern.
- + Accurate spelling for 'Halloween Party'.
- + Strong use of the requested border elements including thorns and webs.
- − Failed to include the required scroll banner and the specific event details (Date, Time, Location).
- − Lacks the 'moody night sky' and cinematic depth requested, appearing more like a 2D illustration.
- − The text is truncated and misses the 'Invitation' part of the title.
Verdict: FLUX.2 [klein] 4B captured the cinematic, vintage gothic aesthetic perfectly and followed all complex layout instructions, though it struggled significantly with text legibility. In contrast, Stable Diffusion 3.5 Large Turbo produced a much simpler 2D-style graphic and failed to include the majority of the required text and the scroll banner. FLUX.2 is the better choice for its superior atmosphere and adherence to the structured elements of the prompt.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent PBR material rendering on the salmon and plate textures.
- + Clean minimalist aesthetic that follows the solid background requirement.
- + Accurate text rendering for the word JAPAN.
- − The text 'SUSHI' is missing the letter 'I'.
- − The flag icon does not represent Japan (resembles Austria or Latvia).
- − Lacks the 'diorama base' requested in the prompt.
Stable Diffusion 3.5 Large Turbo
- + Successfully incorporates the 'diorama base' on a raised wooden platform.
- + Better adherence to the isometric miniature 3D cartoon style requested.
- + Follows the composition requirement with text and flag elements integrated into the scene.
- − Severe spelling error with 'SIIHI' instead of 'SUSHI'.
- − The flag icon is incorrect (resembles Poland).
- − The background is a gradient grey/blue rather than the requested 'solid light blue'.
Verdict: Both models failed to correctly render the Japanese flag or spell 'SUSHI' perfectly, but they excelled in different prompt areas. FLUX.2 [klein] 4B produced a much cleaner, more realistic material finish, while Stable Diffusion 3.5 Large Turbo better understood the 'isometric diorama' and 'top-center' layout requirements. FLUX.2 is slightly better overall due to the high quality of the PBR textures and cleaner typography despite the missing letter.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent adherence to the prompt including all four distinct animal species.
- + Beautiful lighting with visible god rays and backlighting on the fur.
- + High level of detail in the meadow flora and textures.
- − Generated two kittens instead of a baby bunny.
- − The fox's tail is slightly merged with the kitten in the background.
Stable Diffusion 3.5 Large Turbo
- + Strong expressive eyes on all subjects.
- + Clean, vibrant colors that pop against the dark background.
- − Failed to include the four distinct species, showing ambiguous dog/cat hybrids.
- − Lacks the requested 'gold rays' and sunrise atmosphere, feeling more like a studio-lit shot.
- − Missed the bunny and fox kit entirely.
Verdict: FLUX.2 [klein] 4B followed the complex multi-subject prompt much more effectively, managing to render distinct characteristics for the puppy, kittens, and fox, while capturing the atmospheric lighting requested. Stable Diffusion 3.5 Large Turbo struggled with the variety of subjects, producing a simplified image with generic-looking animals that failed to meet most of the prompt's specific requirements.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent minimalist vector style execution
- + Very clean professional layout
- + Accurate representation of the cloche dome
- − Failed the primary text prompt, spelling it 'FLAXTION'
- − Redundant 'Est. 1720' text below the banner
Stable Diffusion 3.5 Large Turbo
- + Captured the correct brand name 'Florin' (with minor typo 'Caffeé')
- + Rich vintage aesthetic with high contrast
- + Good use of the requested banner and steam elements
- − Confused a cloche with a coffee mug, resulting in a strange hybrid object
- − Text rendering on the banner has artifacting on the final 'n'
- − Less minimalist than requested
Verdict: FLUX.2 [klein] 4B produces a much cleaner, more 'logo-ready' vector graphic that perfectly captures the minimalist prompt, but it fails significantly on name spelling. Stable Diffusion 3.5 Large Turbo gets the primary name much closer and follows the vintage vibe well, though it creates a confusing hybrid object by adding a mug handle to the cloche dome. FLUX.2 is the better logo designer, while SD3.5 is the better text follower.
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] 4B
Stable Diffusion 3.5 Large Turbo
AI Judge Analysis
FLUX.2 [klein] 4B
- + Successfully followed the multi-step structural requirement with distinct icons for orbit and descent
- + Clean vector style with consistent iconography that feels like a modern infographic
- + Included specific details requested like the Apollo 11 names and 'Tranquility'
- − Significant spelling errors in almost all text elements
- − Rocket design is generic and doesn't strongly resemble a Saturn V
Stable Diffusion 3.5 Large Turbo
- + Excellent aesthetic value with a dynamic, high-quality vector composition
- + Great adherence to the NASA-inspired color palette and 'clean, modern' style
- + Text is somewhat more legible in titles, though body text is placeholder gibberish
- − Failed the specific 6-step infographic sequence requested in the prompt
- − Missing specific icons requested for step-by-step phases like 'Earth Orbit' or 'Descent'
- − The astronaut on a ladder doesn't fit the specified 'lunar module descending' step
Verdict: FLUX.2 [klein] 4B better adhered to the complex structural requirements of the prompt by attempting all six specific infographic steps, whereas Stable Diffusion 3.5 Large Turbo created a beautiful but more generic poster. While Stable Diffusion 3.5 Large Turbo has superior artistic composition and color usage, FLUX.2 is the winner for functional prompt adherence and following the specific icon instructions.
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