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
#32 of 62 in Text-to-Image
Qwen Image 2512
#30 of 62 in Text-to-Image
Where the votes landed
FLUX.2 [klein] 4B
0%
win rate
Ties
0%
Qwen Image 2512
0%
win rate
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent photorealism with convincing textures on the book and sphere.
- + Accurate adherence to the lighting direction from the left.
- + The plant is clearly visible through the glass as requested.
- − The glass cube has an unusually thick mirrored base that wasn't explicitly requested.
- − The book is slightly too small for the cube's dimensions, looking a bit like a coaster.
Qwen Image 2512
- + Strong composition with better proportions between the book and the cube.
- + Complex and realistic reflections within the glass panels.
- + Great texture on the book's paper edges and cover.
- − The glass has a heavy cyan/teal tint rather than being clear.
- − The plant behind the cube is less distinct through the glass compared to Model A.
Verdict: Both models followed the prompt instructions perfectly, including the specific spatial relationships between objects. FLUX.2 [klein] 4B is preferred for its superior clarity and accurate color of the glass, whereas Qwen Image 2512 introduced a heavy green tint to the glass that distracts from the 'clear cube' concept.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent depiction of raining conditions with visible raindrops and realistic road reflections.
- + Captures the action of repairing the bicycle with a natural, candid pose.
- + Strong rendering of skin texture and clothing materials.
- − The structural anatomy of the bicycle is slightly nonsensical around the chain and seat post area.
- − The framing feels a bit centered despite the request for 'imperfect framing'.
Qwen Image 2512
- + Successfully captures a candid direct gaze with natural facial features.
- + The motion blur on the background cars better matches the prompt's request for movement.
- + The red bicycle color and texture feel authentic to an older, used bike.
- − The subject is posing/looking at the camera rather than 'repairing' the bicycle as requested.
- − The wet ground reflections are somewhat blurry and less defined than in the other model.
- − The rain effect is very subtle to the point of being nearly invisible.
Verdict: FLUX.2 [klein] 4B follows the action of the prompt much better, showing the man actually engaged in repairing the bike with superior environmental effects for rain and reflections. While Qwen Image 2512 produces a very realistic portrait with good motion blur in the background, it fails the primary action of the prompt by having the man sit still and look directly at the lens. FLUX.2 is the overall winner for its atmosphere and adherence to the specific activity described.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent symmetry and clarity in the armor engravings.
- + Very crisp, clean skin texture and anatomically accurate facial features.
- + Realistic lighting interactions on the metal surfaces.
- − The 'battle-worn' appearance feels a bit clean, with aesthetic rather than realistic grime.
- − The hair braids are simple and lack the detailed integration of beads mentioned.
Qwen Image 2512
- + Superb adherence to the 'battle-worn' and 'dirt' descriptors with realistic skin texture.
- + Intricate hair braiding with clearly visible beads and more cinematic lighting.
- + Greater detail on the leather straps and buckle.
- − Slightly more chaotic composition with blurred elements near the edge of the armor.
- − The scars look a bit more like surface scratches than deep battle marks.
Verdict: While FLUX.2 [klein] 4B produces a very clean and symmetrical portrait with impressive armor clarity, Qwen Image 2512 captures the 'battle-worn' and 'lifelike' atmosphere requested in the prompt much more effectively. Qwen's attention to the specific details of the braided hair, beads, and gritty skin textures makes for a more compelling and accurate interpretation.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent photographic quality and realism in the food imagery
- + Clean and professional layout that looks like a high-end menu
- + Good use of white space and typography hierarchy
- − Text rendering is mostly gibberish despite the professional look
- − Section headers do not perfectly match the requested categories
Qwen Image 2512
- + Strong adherence to the 'grid' layout requested
- + Successfully included specific requested sections like 'Appetiizizers'
- + Good use of vibrant color accents in the UI elements
- − Food photos look more generic and less realistic than the competitor
- − Typography is a bit clunky and less legible in the lower sections
Verdict: FLUX.2 [klein] 4B produces a much more realistic and professionally designed menu that better captures the 'modern minimalist' aesthetic, although its text generation is weak. Qwen Image 2512 follows the specific prompt instructions for layout and sections more closely, but the overall visual quality of the food and the design feels more like a template than a high-end product.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent photorealistic rendering of the burger textures
- + Accurate rendering of the secondary text and price
- + Strong use of glow effects and sparks for a dynamic feel
- − Failed the main text prompt with severe spelling errors ('MAAC AGIR BUIRCGRE')
- − Did not create an 'exploded' view; the burger is fully assembled
Qwen Image 2512
- + Perfect adherence to text prompts including the title 'MAGIC BURGER'
- + Accurately represents the 'exploded' view with suspended components
- + Atmospheric use of fire and glowing embers matches the prompt perfectly
- − Minor text omission in the secondary message ('TIME' is missing)
- − The suspended fragments are slightly messy compared to the main burger
Verdict: Qwen Image 2512 is the superior choice because it correctly followed the core creative requirement of an 'exploded' burger and successfully rendered the primary brand name. While FLUX.2 exhibited higher realism in the burger's texture, its failure to separate the components and the gibberish text render it unusable for the specific ad request.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent chalk texture throughout the board
- + Authentic smudging and dust remnants on the chalkboard surface
- + Follows the prompt for a flat, top-down-style view of the menu board
- − Numerous spelling errors including 'Truffel', 'Musheram', 'Ootrpous', and 'Brawn Buter'
- − Inconsistent spacing in the title 'TODAY'S S TPECIALS'
Qwen Image 2512
- + Near-perfect spelling of complex menu items
- + Consistent and aesthetically pleasing elegant cursive handwriting
- + Balanced composition that fills the board effectively
- − The text looks more like a digital brush/font than authentic chalk
- − Minor typo 'Risitto' in the first menu item
Verdict: Qwen Image 2512 is the superior choice because it manages to render almost all the text correctly, including the long third item, whereas FLUX.2 [klein] 4B suffers from significant spelling and spacing issues. While FLUX.2 has better 'chalk' texture and realism in the board's surface, the legibility and accuracy of Qwen Image 2512 make it more usable.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent background detail with a vibrant galaxy and realistic Earth horizon.
- + Physically grounded composition with dust clouds appearing beneath the horse's hooves on the atmosphere.
- + Clear and well-proportioned space suit design.
- − The horse's front right leg has an anatomically awkward bend at the fetlock.
- − The request for 'horse on top' was interpreted as the horse being on top of the atmosphere, rather than the literal inversion of a rider.
Qwen Image 2512
- + Dynamic 'rearing' pose adds to the cinematic quality of the image.
- + Clean lighting on the astronaut's helmet and life support pack.
- + Higher contrast between the subject and the dark space background.
- − Failed to interpret the 'horse on top' logic similarly to Model A, sticking to a conventional rider layout.
- − The astronaut's face inside the helmet appears slightly distorted and unrealistic.
Verdict: Both models failed the negative constraint 'horse on top, not vice versa,' instead generating a standard astronaut riding a horse. FLUX.2 [klein] 4B is the winner due to its superior artistic composition, including the 'astrogust' under the horse's hooves and a more detailed, expansive background compared to Qwen Image 2512.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent photorealism in the textures of the capybara's fur and the woman's coat
- + Accurate depiction of a yellow taxi's interior lighting and perspective
- + High fidelity in rendering the woman's face and phone interaction
- − The steering wheel placement is slightly awkward relative to the driver's seat
- − The capybara's 'paws' look a bit like bird feet/claws
Qwen Image 2512
- + Strong frontal composition that clearly shows both subjects
- + The capybara's hat is more traditionally 'taxi driver' style
- + The bored expression on the passenger is very well captured
- − The passenger is seated on the wrong side relative to the driver in most standard cars
- − The capybara's paws look like strange, leathery human fingers
- − Low-quality rendering of the car's hood and wipers in the foreground
Verdict: FLUX.2 [klein] 4B produces a much more realistic and cohesive image with superior lighting and texture. While Qwen Image 2512 follows the prompt's layout request well, it suffers from anatomical issues with the paws and a lower overall visual quality in the foreground.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent moody atmosphere and textures.
- + Accurate pumpkin glowing effect matches the cinematic lighting request.
- − Significant spelling errors in the title text such as 'Hallbwom'.
- − Incorrect date '30.10.206' and typos in 'Loεton' and 'friqlts'.
Qwen Image 2512
- + Text is much more accurate and legible than the competitor.
- + High quality illustration of the twisted trees and thorny border.
- + Successfully included all event details with very minor spelling issues.
- − Minor spelling error in the title ('Hallowern' instead of Halloween).
- − The scroll banner is slightly overlapping the pumpkin in a busy way.
Verdict: While both models followed the stylistic prompt well, Qwen Image 2512 is the clear winner due to superior text rendering. FLUX.2 [klein] 4B failed significantly on the specific text requirements, with multiple garbled words and incorrect digits in the date, whereas Qwen Image 2512 only had one minor typo in the main title.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent PBR materials with realistic translucency on the fish
- + Clean and minimal composition that adheres to the square format
- + Accurate 45-degree isometric perspective
- − Text is incomplete, displaying 'SUSH' instead of 'SUSHI'
- − The flag icon is incorrect, resembling the flag of Austria rather than Japan
- − Lacks the requested 'diorama base', placing the plate directly on the background
Qwen Image 2512
- + Perfect text rendering for both 'JAPAN' and 'SUSHI'
- + Accurately depicts the Japanese flag icon
- + Follows the 'miniature diorama base' instruction much better with a raised platform
- − Lighting and materials feel slightly more 'plastic' compared to the PBR request
- − The garnish is slightly more cluttered than the 'minimal' request
Verdict: Qwen Image 2512 is the clear winner as it successfully rendered all text and the correct national flag, while also following the specific 'miniature diorama base' prompt. FLUX.2 [klein] 4B produced more realistic textures, but failed significantly on the text/flag accuracy and composition requirements.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Excellent depiction of god rays and sunrise lighting
- + Dynamic and playful composition with animals physically interacting
- + High-quality fur texture and butterfly detail
- − Failed to include the baby bunny
- − Included two kittens instead of one kitten and one bunny
Qwen Image 2512
- + Correctly included all four requested animals (dog, cat, bunny, fox)
- + Impressive sharpness and detail on the animals' faces
- + Adheres well to the 'expressive eyes' part of the prompt
- − The posing is a bit static for a 'playfully chasing and tumbling' prompt
- − The fox's anatomy is slightly forced into the group hug pose
Verdict: Qwen Image 2512 is the winner because it successfully included all four requested animals, including the bunny which FLUX.2 [klein] 4B completely omitted. While FLUX.2 captured a more dynamic 'tumbling' action and better environment lighting, Qwen's superior prompt adherence for the subject matter makes it the better result.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Clean vector aesthetic suitable for a real logo
- + Accurate color palette and subtle parchment texture
- + Includes the accent mark on the 'è' correctly
- − Misspelled name as 'CAFFÈ FLAXTION'
- − Redundant date text appearing twice
- − Steam detail is overly simplistic
Qwen Image 2512
- + Perfect spelling of the restaurant name
- + Beautifully rendered vintage illustration style
- + Dynamic and well-integrated banner for the date
- − Lean towards an illustration rather than a minimalist vector emblem
- − Steam visual is somewhat cluttered
- − Linework on the cloche is quite busy
Verdict: Qwen Image 2512 is the clear winner as it correctly followed the text instructions, accurately spelling 'Caffè Florian', whereas FLUX.2 misspelled the name as 'Flaxtion'. While FLUX.2 achieved a more 'minimalist' vector look, Qwen's superior text rendering and artistic vintage execution make it a more usable and professional asset.
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.”
AI Judge Analysis
FLUX.2 [klein] 4B
- + Clean vector aesthetic with pleasant color blocking
- + Includes names of the astronauts in the supporting details
- + Follows the muted NASA-inspired color palette well
- − Text is largely nonsensical and poorly rendered
- − Iconography is repetitive and does not distinguish well between steps
- − The rocket design is generic and does not resemble the Saturn V
Qwen Image 2512
- + Layout follows the logical progression of mission steps more clearly
- + High quality text rendering for the main title and specific technical terms
- + Iconography for the Lunar Module and Saturn V is more accurate to the mission
- − Some numbering errors in the sequence (two step 2s, two step 3s)
- − Includes 'Steps stop at landing' as literal text on the poster
- − Slightly messy labeling with overlapping text and icons
Verdict: Qwen Image 2512 is the superior choice because it provides a much more recognizable and mission-accurate set of icons, including a proper Saturn V and Lunar Module. While it has some numbering errors and included the prompt instructions as text, FLUX.2 [klein] 4B failed significantly on the specific requested steps and produced largely garbled text.
Explore each model
Improved version of Alibaba's Qwen image model with better text rendering, finer natural textures, and more realistic human generation.