Black Forest Labs' open-weights image generation model with frontier performance, available for non-commercial local deployment
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
FLUX.2 [dev]
24.3
arena score
#19 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Qwen Image 2512
22.9
arena score
#30 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.2 [dev]
83.3%
win rate
Ties
0.0%
Qwen Image 2512
16.7%
win rate
83.3%
0.0% ties
16.7%
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 [dev]
Qwen Image 2512
67% wins
0% ties
33% wins
AI Judge Analysis
FLUX.2 [dev]
- + Excellent rendering of caustic light and refractions through the glass cube.
- + The blue sphere looks like real glass/crystal, matching the overall aesthetic.
- + Very soft, natural window lighting that creates a photorealistic atmosphere.
- − The blue sphere appears to be floating unnaturally within the cube.
- − The glass cube has rounded, thick edges that make it look a bit like a construction block rather than a hollow container.
Qwen Image 2512
- + The physics are more grounded, with the sphere resting realistically on the bottom of the cube.
- + The glass cube has sharp, defined edges and clear internal reflections.
- + Excellent texture work on the book cover and the wood grain of the table.
- − The blue sphere has a matte, plastic texture that looks a bit less premium than the glass surroundings.
- − The plant in the background is slightly more obscured compared to Model A.
Verdict: Both models followed the complex spatial prompt perfectly, including the specific placement of the plant behind the glass. FLUX.2 [dev] produces a more artistic, luminous image with beautiful caustics, but the sphere is floating; Qwen Image 2512 produces a more grounded, realistic scene with better material definition for the book and table.
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 [dev]
Qwen Image 2512
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [dev]
- + Excellent execution of the motion blur request for passing cars.
- + Highly realistic skin and hair textures on the subject.
- + Successfully captures the action of 'repairing' the bicycle with complex hand positioning.
- − The structural integrity of the bicycle becomes messy and nonsensical in the lower section.
- − There are some minor digital artifacts in the blurred cars that look a bit painterly.
Qwen Image 2512
- + Very coherent bicycle structure and realistic mechanical details.
- + Follows the prompt for an elderly Japanese man with natural skin textures.
- + Good use of wet pavement reflections and shallow depth of field.
- − Failed to include 'motion blur' from passing cars; the background traffic looks mostly stationary.
- − The subject is posing/looking at the camera, which contradicts the 'candid' part of the prompt.
- − The subject is just crouching by the bike rather than 'repairing' it.
Verdict: FLUX.2 [dev] followed the technical instructions for motion blur and the candid, active nature of the scene much better than Qwen Image 2512. While Qwen Image 2512 rendered a more structurally sound bicycle, FLUX.2 [dev] achieved a more convincing cinematic street photography aesthetic with genuine movement and a candid feel.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent depiction of ornate engraving on the pauldrons.
- + Clear, colorful beads integrated into the hair braids.
- + Strong bokeh effect with distinct sparks that enhance the atmosphere.
- − The facial scars look a bit like surface paint rather than deep, weathered wounds.
- − The overall image feels a bit too clean for a 'battle-worn' paladin.
Qwen Image 2512
- + Highly realistic skin texture with grime and more convincing scars.
- + Superb catchlights in the eyes creating a lifelike gaze.
- + Better lighting integration where the torchlight naturally highlights the ridges of the armor.
- − The armor engraving is slightly less intricate and detailed compared to the other model.
- − The beads in the hair are smaller and less prominent than requested.
Verdict: Both models followed the prompt exceptionally well, but Qwen Image 2512 wins due to its superior character depth and realistic textures, capturing a truly 'battle-worn' look. While FLUX.2 [dev] produced beautiful armor engravings and distinct beads, it felt slightly more sanitized and staged compared to the gritty realism of Qwen Image 2512.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent typography with clean, professional sans-serif fonts
- + Highly realistic food photography that looks appetizing and professional
- + Perfectly followed the section requirements for Appetizers, Pizza, and Mains
- − Text contains gibberish/nonsense words despite good formatting
- − Some layout elements overlap slightly in the pizza section
Qwen Image 2512
- + Strong use of vibrant color accents in the layout
- + Excellent grid organization of food photos at the top
- + Clean overall aesthetic for a casual dining flyer
- − Typographical errors in main headings like 'APPETIIZIZERS' and 'MEANS'
- − The 'Mains' category was partially merged with appetizers and mislabeled
- − Food photography is slightly lower in texture quality compared to Model A
Verdict: FLUX.2 [dev] produces a significantly more professional layout that resembles a real restaurant menu, featuring high-quality food photography and clear sectioning for all requested categories. While Qwen Image 2512 has a nice punchy color palette, it suffers from spelling errors in large titles and fails to clearly distinguish between 'Pizza' and 'Mains' in the text sections.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent typography with a cohesive neon fiery glow.
- + Clean composition that feels like a professional marketing ad.
- + Perfect adherence to all requested text elements, including the currency symbol.
- − The 'exploded' effect is a bit static compared to Model B.
- − The bun texture is slightly smooth/artificial.
Qwen Image 2512
- + Superb photorealistic detail in the meat patty and fresh toppings.
- + Dynamic sense of motion with flying debris and sauce splatters.
- + Rich, high-contrast lighting that fits the 'fiery' theme perfectly.
- − Missed the word 'TIME' in the secondary 'LIMITED TIME ONLY' message.
- − The starburst design is slightly more cluttered than Image A.
Verdict: FLUX.2 [dev] followed the text instructions perfectly, including every word and the specific starburst request, resulting in a cleaner ad layout. However, Qwen Image 2512 provided a much more realistic and dynamic image of the food itself, though it failed to correctly render the phrase 'LIMITED TIME ONLY' by omitting a word.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent chalk texture throughout the board
- + Perfect spelling on all requested items
- + Realistic board smudges and background integration
- − Alignment is slightly inconsistent with 'with' and '$28' floating awkwardly
- − The cursive requested for the title is more of a print-script blend
Qwen Image 2512
- + Beautiful elegant cursive for the header
- + Better spacing and layout of the menu items
- + Consistent handwriting style across all sections
- − Spelling error in 'Risitto' instead of 'Risotto'
- − The text lacks the granular chalk grain texture, appearing a bit too smooth or digital
Verdict: FLUX.2 [dev] excels in realism with incredible chalk physics and texture, and it captures all text without spelling errors. Qwen Image 2512 has better aesthetic cursive and cleaner layout, but fails on the spelling of 'Risotto'.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.2 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + High visual appeal with a cinematic, glowing galaxy background.
- + Excellent composition using depth of field and atmospheric lighting from the planet below.
- + Very high resolution with fine details in the horse's mane and the space suit textures.
- − Failed the specific spatial instruction for the horse to be on top of the astronaut.
- − The horse's back hooves are partially obscured by an odd smoky effect.
Qwen Image 2512
- + Clean, sharp lighting on the astronaut and horse anatomy.
- + Good rendering of the face through the visor without distortion.
- + Highly detailed leather tack and saddle equipment.
- − Failed the specific spatial instruction for the horse to be on top of the astronaut.
- − The composition feels slightly flatter and less cinematic compared to the other model.
- − The horse's anatomy becomes slightly disconnected where the front legs meet the chest.
Verdict: Both models failed the negative/inverse constraint to place the horse on top of the astronaut, instead defaulting to the common trope of an astronaut riding a horse. FLUX.2 [dev] is the superior image due to its more immersive cinematic lighting, better integration of the subject into the space environment, and more natural textures.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent photorealistic texture on the capybara fur and jacket.
- + The composition feels more natural and cinematic.
- + Better adherence to the 'bored' expression for the passenger.
- − The passenger is sitting in the front seat or middle, which is less common for a taxi passenger.
- − The capybara paws look slightly more like humanoid hands with claws.
Qwen Image 2512
- + Correctly places the passenger in the back seat behind the driver.
- + The capybara's professional pose and symmetrical gaze are quite striking.
- + Good attention to the taxi hat detail.
- − The passenger's expression looks sad or distressed rather than 'normal and bored'.
- − The perspective of the steering wheel and paws is a bit distorted.
- − The background lighting is less integrated with the foreground.
Verdict: FLUX.2 [dev] produces a more photorealistic image with superior textures and a more convincing 'bored' expression on the passenger, though it places the passenger in the front seat. Qwen Image 2512 correctly places the woman in the back seat but fails to capture the casual/bored vibe requested, instead portraying a sad expression.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Perfect spelling of all text, including 'Halloween'.
- + Clean, professional typographical layout with excellent legibility.
- + Very high-quality border consisting of intricate thorns and webs that feels dark and gothic.
- − The central jack-o-lantern is quite dark/black, which makes it blend into the background slightly more than desired.
Qwen Image 2512
- + Vibrant, classic orange jack-o-lantern with a strong glow effect.
- + Dynamic composition with more prominent bats and detailed twisted trees.
- + The banner scroll is stylistically integrated well into the gothic theme.
- − Contains a spelling error in the main title: 'Hallowern'.
- − The font choice for the body text at the bottom is less stylized and 'gothic' compared to the title.
Verdict: FLUX.2 [dev] followed the text instructions perfectly, delivering an error-free invitation with a sophisticated gothic aesthetic and superior border detail. Qwen Image 2512 created a more visually striking central pumpkin and trees, but failed on the critical title text by misspelling 'Halloween' as 'Hallowern'.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent adherence to the 'raised diorama base' with very clean geometric lines.
- + Perfect execution of 'small flag icon' and requested text hierarchy.
- + Superior lighting and refined textures that feel truly isometric and 3D.
- − The sushi models are a bit simple compared to the high-detail text.
Qwen Image 2512
- + Beautifully detailed and appetizing sushi textures with realistic grain and shine.
- + Playful and cohesive font choice that matches the cartoon-style prompt.
- + Dynamic composition with more interesting garnish elements.
- − Missed the small flag icon requirement, placing it inside the word 'SUSHI' box instead.
- − The base feels a bit more organic/lumpy rather than a 'clean' diorama base requested.
Verdict: FLUX.2 [dev] followed the specific icon and layout instructions more precisely, resulting in a cleaner, more professional-looking isometric graphic. While Qwen Image 2512 produced more visually rich and detailed sushi models, it failed to include the flag as a separate icon and the base was less refined than the one in the FLUX.2 [dev] output.
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 [dev]
Qwen Image 2512
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [dev]
- + Excellent depiction of god rays and sunrise lighting
- + Very natural fur texture and backlighting
- + Includes multiple bunnies as requested by the plural prompt
- − The animals are sitting still rather than 'chasing and tumbling' as requested
- − One of the butterflies has slightly distorted wings
Qwen Image 2512
- + Strong composition with a central focus
- + Clear, expressive eyes and sharp details on all four animal types
- + Vibrant colors and well-defined butterflies
- − The anatomy where the puppy's paws meet the other animals is a bit muddled
- − Lacks the 'tumbling' action requested in the prompt
- − God rays are a bit more artificial/stylized compared to Model A
Verdict: Both models followed the prompt well, but FLUX.2 [dev] produced a more photorealistic scene with superior lighting and atmospheric effects like dew sparkles and natural god rays. Qwen Image 2512 has a charming composition but suffers from slight anatomical blending where the animals overlap and lacks the atmospheric depth of FLUX.2.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent adherence to the minimalist, vector-style logo request
- + Perfectly rendered text including the specific accent mark on 'Caffè'
- + Very clean and balanced composition suitable for actual branding
- − The steam element is a bit small and simplified compared to the rest of the icon
Qwen Image 2512
- + Beautifully detailed woodcut-style illustration of the cloche dome
- + Dynamic and artistic steam effects
- + Rich, warm tonal palette that matches the 'vintage' request
- − Violates the 'minimalist' and 'vector emblem' constraints with excessive detail
- − The banner and text layout is slightly unbalanced and more of an illustration than a logo
Verdict: FLUX.2 [dev] followed the technical requirements of the prompt much more accurately, producing a true minimalist vector-style logo with flawless typography. While Qwen Image 2512 created a visually stunning vintage illustration, it ignored the 'minimalist' constraint and produced a complex piece that would be difficult to use as a functional logo.
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 [dev]
Qwen Image 2512
AI Judge Analysis
FLUX.2 [dev]
- + Excellent adherence to the color palette requesting navy, white, muted red, and light gray.
- + Individual icons are very high quality with crisp lines and a consistent modern vector style.
- + Correctly identifies and creates icons for all requested mission phases.
- − The layout is cluttered with 9 icons instead of follow the requested 6-step sequence.
- − Text rendering contains high gibberish for most category labels.
- − Does not use a logical flow/path for the icons, making it more of a gallery than an infographic.
Qwen Image 2512
- + Follows the numbered 6-step sequence more logically in a vertical flow.
- + Features better text legibility for the main title and most step titles.
- + Accurately represents the Saturn V and Lunar Module with technical detail in a vector style.
- − Includes redundant labels (two labels for Earth Orbit) and incorrect numbering.
- − The palette leans a bit more into gold/orange than the requested muted red/light gray NASA palette.
- − The font choice and white drop shadows feel slightly less modern than the requested style.
Verdict: Both models struggled with the specific 6-step sequence, but FLUX.2 [dev] produced significantly higher-quality individual vector illustrations that perfectly matched the requested NASA color scheme. Qwen Image 2512 had better text legibility and a more logical infographic flow, but the redundant labels and slightly muddy colors make it less visually appealing than FLUX.2 [dev].
Next steps
Explore each model
Improved version of Alibaba's Qwen image model with better text rendering, finer natural textures, and more realistic human generation.