Black Forest Labs' open-weights multimodal flow transformer for in-context image generation and editing, available for non-commercial use with character consistency and style transfer capabilities
Settled by community votes across 17 shared challenges, with an AI judge weighing in on each.
FLUX.1 Kontext [dev]
#54 of 62 in Text-to-Image
FLUX.2 [dev]
#19 of 62 in Text-to-Image
Where the votes landed
FLUX.1 Kontext [dev]
0%
win rate
Ties
0%
FLUX.2 [dev]
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.1 Kontext [dev]
- + Excellent surface textures, particularly on the red book cover and wooden table grain.
- + Strong adherence to the 'small' descriptor for the sphere within the cube.
- + High resolution with clean, sharp edges on the glass cube.
- − The glass cube lacks a front face, appearing more like a glass stand or table than a sealed cube.
FLUX.2 [dev]
- + The cube is rendered with all six sides clearly defined, showing realistic thickness in the glass panels.
- + Highly accurate lighting, with the blue sphere casting a clear reflection on the bottom glass panel.
- + Excellent depth of field with the plant correctly positioned behind the glass.
- − The blue sphere is slightly off-center, which might be a minor compositional drawback for some.
Verdict: Both models followed the prompt perfectly, capturing all objects and the specific lighting conditions. FLUX.2 [dev] is the winner because it successfully rendered a complete, three-dimensional glass cube with realistic panel thickness, whereas FLUX.1 Kontext [dev] produced a shape that appears to have no front glass face.
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.1 Kontext [dev]
- + Clean composition with vibrant colors
- + Good representation of a red bicycle and wet pavement reflections
- − Subject is just sitting on the bike rather than repairing it
- − Cars in the background are stationary with headlights on, missing the requested 'motion blur'
- − The rain appears as static, uniform lines across the entire frame
FLUX.2 [dev]
- + Successfully captures the man actively 'repairing' the bicycle as requested
- + Excellent execution of motion blur on passing cars in the background
- + Realistic skin texture and 'imperfect framing' create a high-quality candid feel
- − The secondary handlebars in the foreground are a bit cluttered/confusing structurally
- − The framing is very tight, cutting off the bottom of the bicycle
Verdict: FLUX.2 [dev] followed the complex prompt instructions much more accurately than FLUX.1 Kontext [dev], particularly the 'motion blur' and 'repairing' aspects. While FLUX.1 Kontext [dev] produced a clean image, it failed several key prompt requirements, whereas FLUX.2 [dev] delivered a much more realistic, cinematic, and candid atmosphere that matched the 50mm lens and street photography aesthetic.
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.1 Kontext [dev]
- + Excellent dramatic lighting and high-contrast composition
- + Highly intricate engraving patterns on the breastplate
- + Realistic skin texture and lifelike eyes
- − Missed the request for braided hair with beads
- − The battle-worn attributes (scars/dirt) are very subtle compared to the prompt
FLUX.2 [dev]
- + Perfect adherence to specific details like braided hair with beads
- + Excellent rendering of weathered leather straps and textured cloth
- + Stronger 'battle-worn' aesthetic with visible scars and dirt
- − Lighting is a bit flat across the face compared to Model A
- − Metal texture looks slightly more like brushed steel rather than ornate gold/bronze plate
Verdict: While FLUX.1 Kontext [dev] creates a more striking cinematic image, FLUX.2 [dev] is the superior model for prompt adherence, accurately including the difficult details like the braided hair with beads and the specific layered textures. FLUX.2 [dev] also interprets 'battle-worn' much more effectively through the visible facial scarring and grime.
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.1 Kontext [dev]
- + Features a truly minimalist grid layout that is very modern.
- + The food photos are distinct, high-quality, and colorful.
- + Text is bold and follows the requested sans-serif style.
- − Fails to include the 'Pizza' or 'Mains' labels requested in the prompt.
- − The text is largely nonsensical gibberish.
- − The layout is more of a graphic poster than a functional restaurant menu.
FLUX.2 [dev]
- + Excellent adherence to the specific section requests (Appetizers, Pizza, Mains).
- + Includes realistic menu elements like pricing and descriptions.
- + The layout feels more like a functional two-page menu.
- − Text rendering is messy with many character artifacts and misspellings.
- − The pizzas in the 'Appetizer' section create a logical inconsistency.
- − The color gradients on the accents feel a bit dated compared to Model A.
Verdict: Model B (FLUX.2 [dev]) is the winner because it successfully incorporated all three requested menu sections (Appetizers, Pizza, Mains) and included pricing, making it look like a real menu. Model A (FLUX.1 Kontext [dev]) has a cleaner aesthetic but failed to include the specific categorizations requested, resulting in a design that looks more like a magazine layout than a restaurant 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.”
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent typography rendering for the primary title
- + High-quality, appetizing burger rendering
- + Strong environmental lighting from the coals below
- − Failed the 'exploded' portion of the prompt by keeping the burger nearly fully assembled
- − Contains a prominent spelling error in the word 'ONLY' (rendered as 'LNHLY')
- − The price starburst looks like a flat graphic compared to the rest of the scene
FLUX.2 [dev]
- + Perfectly executed the 'exploded' burger request with deconstructed layers
- + All text is spelled correctly and rendered with the requested fiery, glowing effect
- + Dynamic food photography feel with sauce droplets suspended in air
- − The bun texture is slightly less detailed compared to Model A
- − The lettuce looks a bit less realistic than the other components
Verdict: FLUX.2 [dev] significantly outperformed FLUX.1 Kontext [dev] by correctly interpreting the complex 'exploded' layout and ensuring all text was spelled accurately. While FLUX.1 Kontext produced a high-quality static image, it failed the core composition requirement and introduced a major typo in the secondary text.
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.1 Kontext [dev]
- + Excellent chalk-like texture on the 'TODAY SPECIALS' text
- + Clean framing and lighting for a menu board
- − Numerous spelling and grammar errors including 'Mushroom Mashroom' and 'Risoktso'
- − The date is rendered as illegible gibberish
- − Text duplication on the second menu item line
FLUX.2 [dev]
- + Perfect spelling on almost every item including the date and the specific menu items
- + Very realistic chalk dust and smearing effects on the chalkboard background
- + Superior adherence to the specific 'Brown Butter' prompt completion
- − The 'with' on the second line is slightly muddy/smudged
- − The text doesn't quite fill the board symmetrically
Verdict: FLUX.2 [dev] significantly outperformed FLUX.1 Kontext [dev] in terms of text accuracy, spelling, and adherence to the prompt details. While FLUX.1 Kontext struggled with basic spelling and generated gibberish for the date, FLUX.2 [dev] correctly identified all items and rendered them in a more authentic chalk style with natural smeared textures.
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.1 Kontext [dev]
- + Excellent adherence to the specific 'horse on top' spatial instruction
- + Clear and anatomically correct facial features through the helmet
- + Clean lighting and high-quality textures on the space suit
- − The horse's legs are awkwardly positioned and cut into the astronaut's space
- − Background elements are somewhat sparse and less cinematic than the competitor
FLUX.2 [dev]
- + Beautiful cinematic lighting and rich background detail including a nebula and moon
- + High quality textures on the horse's fur and the complex space suit hardware
- − Completely failed the negative constraint to have the horse on top
- − The horse's front legs have anatomical issues and strange wrapping
Verdict: FLUX.1 Kontext [dev] is the winner because it successfully followed the difficult spatial instruction to place the horse on top of the astronaut, whereas FLUX.2 [dev] ignored this part of the prompt and produced a standard 'astronaut riding a horse' image. While FLUX.2 [dev] had a more visually stunning cinematic background, its failure to adhere to the core prompt logic makes it unsuccessful in this challenge.
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.1 Kontext [dev]
- + Excellent photographic quality and texture on the capybara fur.
- + The lighting inside the cabin feels cinematic and realistic.
- + Successful high-contrast night scene with blurred city lights.
- − The capybara only has one hand on the wheel and it looks like a strange hybrid hand.
- − The capybara's face looks slightly more like a bear or groundhog than a typical capybara.
- − The businesswoman appears to be sitting directly behind the driver rather than for optimal viewing of her and the phone.
FLUX.2 [dev]
- + Captures a more accurate capybara facial structure and head shape.
- + Strictly follows the prompt of 'both front paws on the steering wheel'.
- + Excellent composition showing both the driver and the passenger through the front windshield.
- − The paws on the steering wheel look more like human-primate hands than capybara paws.
- − The lighting is a bit flat and less 'night-like' compared to the other model.
- − The background bokeh is a bit busy and distracting.
Verdict: FLUX.1 Kontext [dev] produced a more atmospheric and high-quality image, but failed on the specific requirement of having both paws on the steering wheel. FLUX.2 [dev] followed the technical instructions of the prompt more closely, including the paw placement and the businesswoman's bored interaction with her phone, although the overall image quality is slightly less refined than the first.
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.1 Kontext [dev]
- + Clear, legible main title text
- + Vibrant jack-o-lantern glowing effect
- + Accurate date and time representation
- − Small banner text is scrambled and illegible
- − The location name 'The Arches' is misspelled as 'ARGIIVAH'S'
- − Lacks the requested 'dark parchment' and vintage atmospheric depth
FLUX.2 [dev]
- + Perfect text rendering for all fields including the small scroll banner
- + Excellent gothic atmosphere with parchment texture, thorns, and twisted trees
- + Fulfills all prompt elements including specific location and cinematic lighting
- − The jack-o-lantern is quite dark/black, which may slightly reduce the 'glowing' contrast
- − Slightly less 'vibrant' color palette compared to Model A
Verdict: FLUX.2 [dev] significantly outperforms FLUX.1 Kontext [dev] by accurately rendering all requested text, including the location and the small scroll banner which Model A failed to spell correctly. Additionally, FLUX.2 better captured the requested 'vintage gothic' and 'parchment' aesthetic, creating a more cohesive and professional invitation design.
Bald man challenge
Image Editing“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Successfully added a thick head of hair that follows the requested prompt.
- + Maintained the overall aesthetic and color palette of the clothing and background.
- − Failed to preserve original facial features, significantly altering the man's face to look younger and different.
- − Modified the style and color of the eyeglasses from the source image.
FLUX.2 [dev]
- + Excellent source preservation, keeping the original face, glasses, and background entirely intact.
- + Highly realistic hair texture with natural lighting that matches the scene.
- − The choice of an afro hairstyle, while full and thick, may feel less 'natural' for this specific individual's existing beard texture and ethnicity compared to other styles.
Verdict: FLUX.2 [dev] performed significantly better as an image editing tool by perfectly preserving the subject's face, glasses, and environment while adding realistic hair. FLUX.1 Kontext [dev] failed the source preservation criteria by completely generating a new face and set of glasses that only vaguely resemble the original person.
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.1 Kontext [dev]
- + Excellent typography style that matches the bold, cartoon aesthetic.
- + High-clarity surfaces and clean geometric shapes.
- + Vibrant color palette that pops against the blue background.
- − The flag icon is incorrect and unrecognizable as the Japanese flag.
- − The sushi design is oversimplified, looking more like an abstract pill than food.
- − Missing the 45-degree angle requested, opting for a flatter perspective.
FLUX.2 [dev]
- + Perfect adherence to the 45-degree isometric projection and diorama base request.
- + High-quality realistic PBR textures on the rice, fish, and wasabi.
- + Accurate Japanese flag icon and excellent spatial distribution of elements.
- − The text is slightly thin compared to the 'large bold' request.
- − Slightly less 'cartoonish' than Model A, though arguably more detailed.
Verdict: FLUX.2 [dev] is the clear winner as it followed every technical instruction, including the isometric 45-degree angle, the specific layout of the diorama, and the correct Japanese flag. While FLUX.1 Kontext [dev] had appealing bold typography, its failure to generate a recognizable flag and its overly abstract sushi render made it less effective than the highly detailed and accurate output of FLUX.2 [dev].
Over-the-top cartoon caricature
Editing“Create a caricature of me and my job. Make it exaggerated and humorous, incorporating my profession as a tv show anchor and my love for dogs and hockey.”
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Strong resemblance to the original subject's facial features and hair.
- + Distinctive vibrant comic-book art style.
- + Effectively incorporates the TV anchor role with the background television.
- − Completely missed the 'hockey' instruction.
- − The dog is rendered as a stylized cartoon/logo rather than a full caricature character.
- − Text rendering is nonsensical and distracting.
FLUX.2 [dev]
- + Successfully incorporates all prompt elements including TV anchoring, dogs, and hockey in a single scene.
- + Authentic caricature style with the classic 'large head, small body' proportions.
- + The setting on an ice rink perfectly blends the requested hobbies with the profession.
- − The facial resemblance to the original person is slightly weaker than in the other model.
- − Unnecessary addition of a secondary male character in the background.
Verdict: FLUX.2 [dev] successfully followed the entire prompt, creating a cohesive scene that combined the subject's career as an anchor with dogs and hockey in a traditional caricature style. FLUX.1 Kontext [dev] captured the subject's likeness better but failed to include the hockey element and used a more generic pop-art style rather than a true caricature.
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.1 Kontext [dev]
- + Excellent sense of motion and playfulness
- + Very clean, bright lighting with a joyful atmosphere
- + Consistent art style across all characters
- − Failed to include the bunny and the red fox kit
- − Included two cats instead of the specific animal variety requested
- − Visual style feels slightly more digital/illustrative than hyper-photorealistic
FLUX.2 [dev]
- + Included all requested animals: puppy, tabby kitten, bunny, and red fox kit
- + Superior texture detail in the fur and environment
- + Excellent realization of god rays and dew sparkles in the lighting
- − The animals are mostly sitting still rather than 'chasing and tumbling'
- − The bunny appears to be duplicated or merging with a second rabbit
Verdict: FLUX.2 [dev] is the clear winner for prompt adherence, as it successfully included all four distinct species requested whereas FLUX.1 Kontext [dev] only generated a dog and two cats. FLUX.2 [dev] also achieved a more convincing hyper-photorealistic look with better environmental effects like dew and volumetric lighting, despite the minor anatomical glitch with the rabbits.
Studio Ghibli Anime Style
Editing“Transform this photo into a Studio Ghibli–inspired illustration. Use soft pastel colors, hand-painted textures, gentle lighting, dreamy backgrounds, and a warm, nostalgic mood”
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent preservation of the original street background and clothing details.
- + Accurately replicates the character poses and spatial layout.
- + Strong line work consistent with modern anime styles.
- − The art style is generic modern anime rather than the specific 'hand-painted' Studio Ghibli aesthetic requested.
- − Colors are fairly saturated rather than soft pastels.
FLUX.2 [dev]
- + Perfectly captures the Studio Ghibli aesthetic with hand-painted textures and soft pastel colors.
- + Effectively creates a 'dreamy background' using flowers and soft clouds while keeping the subject recognizable.
- + Maintains character expressions and the core meme composition.
- − Replaces the background entirely instead of transforming the existing street scene.
- − The man's hand on the right is slightly malformed.
Verdict: FLUX.1 Kontext [dev] did a better job of preserving the source image context by keeping the city street, but the style is too crisp and digital for a Ghibli prompt. FLUX.2 [dev] significantly better adhered to the artistic requirements of the prompt, delivering a beautiful, hand-painted look with a warm, nostalgic mood that perfectly embodies the Studio Ghibli style.
Golden Hour Stroll
Image Editing“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent source preservation, maintaining identical character features and background details.
- + Subtle and realistic hair movement that matches the breeze.
- − The 'flying leaves' are barely noticeable with only a few small green specs present.
- − The overall feel is less 'dynamic' than requested.
FLUX.2 [dev]
- + Strong adherence to the 'energetic and lively' request with significant leaf movement.
- + Dramatic wind-blown hair effect that effectively conveys motion.
- − Struggles with source preservation as the woman's facial features have visibly changed.
- − The leaves appear somewhat flat and lack motion blur compared to the sharp background.
Verdict: FLUX.1 Kontext [dev] did an excellent job of keeping the image exactly as it was while adding subtle edits, but it failed to fully capture the energetic feel of the prompt. FLUX.2 [dev] followed the creative instructions much more effectively with dramatic hair and leaf effects, though it failed to preserve the woman's original face. FLUX.2 [dev] is the winner for better following the core 'dynamic motion' instruction despite the loss in likeness.
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.1 Kontext [dev]
- + Clean vector lines and high-contrast typography
- + Accurate spelling including the grave accent on 'Caffè'
- + Strong minimalist aesthetic that is easily scalable
- − Failed to include the requested banner element
- − The steam effect is overly stylized and lacks clarity
- − The cloche looks more like a dome or architectural piece than a food cover
FLUX.2 [dev]
- + Excellent adherence to all prompt elements including the banner and steam
- + Beautiful vintage texture and paper grain effect
- + Sophisticated typography and professional logo composition
- − The accent mark on 'Caffè' is slightly misplaced/merged with the letter
- − Smaller text size makes 'Est. 1720' harder to read at low resolutions
Verdict: FLUX.2 [dev] followed the prompt much more accurately by including the requested banner and incorporating a much more recognizable cloche and steam icon. While FLUX.1 Kontext [dev] produced a very clean vector, it missed key thematic elements and chose a font that feels less 'vintage' than the requested style.
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.1 Kontext [dev]
- + Stronger adherence to the flat-vector aesthetic requested in the prompt.
- + Uses the specified navy, white, and muted red color palette correctly.
- − Severely garbled text and typos in the headline (e.g., 'APOLO').
- − Icons are abstract and messy, failing to clearly represent the specific mission stages.
- − The layout is disorganized with overlapping text and icons.
FLUX.2 [dev]
- + Excellent visual clarity and professional infographic layout.
- + High-quality iconography that clearly depicts the Saturn V, Lunar Module, and trajectories.
- + Mostly legible text including the names of the crew members.
- − Includes several extra steps not requested in the 6-step prompt, leading to some repetition.
- − Uses a more detailed illustrative style rather than the requested 'flat-vector' style.
- − Some minor spelling artifacts in the smaller labels.
Verdict: FLUX.2 [dev] is the clear winner as it successfully creates a coherent, educational infographic with high-quality icons that match the mission stages. FLUX.1 Kontext [dev] followed the 'flat' style better but failed fundamentally on text rendering, spelling, and compositional clarity.
Explore each model
Black Forest Labs' open-weights image generation model with frontier performance, available for non-commercial local deployment