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 15 shared challenges, with an AI judge weighing in on each.
FLUX.1 Kontext [dev]
17.1
arena score
#54 of 62 in Text-to-Image
Skill signature
· Text-to-Image
OmniGen v2
16.9
arena score
#57 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.1 Kontext [dev]
100.0%
win rate
Ties
0.0%
OmniGen v2
0.0%
win rate
100.0%
0.0% ties
0.0%
Shared challenges
15
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent adherence to lighting instructions with a clear soft window light from the left.
- + Highly realistic textures on the wood grain, book cover, and glass reflections.
- + Accurate spatial placement of all objects including the plant visibility.
- − The glass cube has a mirrored base which wasn't specifically requested but adds visual complexity.
OmniGen v2
- + Successfully includes all objects in the requested spatial relationship.
- + Clean, minimalist composition with logical shadows.
- − The glass cube appears to have thicker, almost liquid-like edges at the top.
- − Lighting feels slightly more flat and less directional compared to Model A.
Verdict: Both models followed the complex spatial prompt perfectly. FLUX.1 Kontext [dev] is the winner due to its superior rendering of textures and more convincing natural lighting from the window, whereas OmniGen v2 has slightly less realistic glass refractive properties and a more generic appearance.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent handling of complex lighting and traffic background
- + High-quality skin texture and realistic clothing fabric
- + Good adherence to the 'reflections on wet pavement' and 'light rain' prompts
- − The man is straddling/starting to ride the bike rather than actively 'repairing' it
- − The composition feels a bit staged with the subject positioned directly in the center of a busy road
OmniGen v2
- + Captures a more authentic 'repairing' stance with the man leaning over the bike
- + Beautifully rendered reflections on the wet asphalt
- + Naturally soft color palette that feels cinematic
- − Significant anatomical and mechanical errors such as the man's leg disappearing into the bicycle frame
- − The hands on the handlebars are warped and lack detail
- − Cars in the background are static rather than showing the requested 'motion blur'
Verdict: FLUX.1 Kontext [dev] produces a much more coherent and technically sound image with superior skin textures and realistic environments, though it misses the specific action of 'repairing'. OmniGen v2 captures the intended mood and pose more accurately but suffers from severe structural artifacts where the man's body merges with the bicycle frame.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent execution of warm torchlight reflecting off the metal armor.
- + Superior detail on the intricate engravings of the plate armor.
- + Convincingly 'battle-worn' appearance with realistic skin texture and scars.
- − Missed the instruction for hair braided with beads.
- − The skin texture looks slightly more digital/painterly compared to the other model.
OmniGen v2
- + Perfect adherence to specific details like braided hair with beads.
- + Very lifelike eye rendering and clear, high-contrast bokeh sparks.
- + Includes the requested leather straps and cloth underlayer with clear textures.
- − The character looks too pristine and clean for a 'battle-worn' description.
- − The dirt on the face looks like stamp-on decals rather than natural grime.
Verdict: OmniGen v2 successfully captured every specific element of the prompt, including the braided hair and beads which FLUX.1 Kontext [dev] missed. However, FLUX.1 Kontext [dev] better captured the 'battle-worn' atmosphere and the complex metallic reflections of the torchlight. OmniGen v2 is the winner for prompt adherence, while FLUX.1 Kontext [dev] has a more cohesive cinematic mood.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent photographic quality and food lighting
- + Clean, high-contrast grid composition
- + Bold, readable typography that feels modern
- − Text is largely nonsensical gibberish
- − Layout is more of a graphic poster than a functional multi-section menu
OmniGen v2
- + Clear hierarchical sections for different food categories
- + More vibrant use of color accents with the background blocks
- + Better adherence to the 'casual dining' professional layout
- − Visual quality of photos is lower and slightly more artificial
- − Text contains multiple misspellings like 'PIZZZZAN' and 'RESTAURATED'
- − Font choice feels less 'modern minimalist' than the prompt requested
Verdict: FLUX.1 Kontext [dev] produces much higher quality food photography and a cleaner aesthetic, but it fails to create a functional menu structure. OmniGen v2 successfully organizes the content into logical categories as requested, though the image quality is lower and the text rendering is significantly less polished. FLUX.1 Kontext [dev] is the preferred choice for its superior visual clarity and professional design feel.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent photographic texture on the meat patty and fresh lettuce.
- + Captures a high-quality atmospheric background with realistic glowing embers and fire.
- + Correctly includes the Euro currency symbol as requested.
- − Failed the core 'exploded burger' prompt instruction, showing a fully assembled burger.
- − Minor typo in the secondary text ('LNHLY' instead of 'ONLY').
OmniGen v2
- + Strong graphic design layout that feels like a completed advertisement.
- + Highly vibrant and cohesive fiery color palette throughout the image.
- + Clean text rendering for the primary 'MAGIC BURGER' title.
- − Failed the 'exploded burger' instruction, showing a static, stacked burger.
- − Missing the Euro currency symbol requested in the prompt.
- − The 'LIMITED TIME ONLY' text is cut off on the right side.
Verdict: Both models failed to deliver the 'exploded' view where components are suspended in mid-air, instead providing standard stacked burgers. FLUX.1 Kontext [dev] is the superior image due to its more realistic textures and adherence to the specific currency requested, whereas OmniGen v2 has significant text clipping and a more plastic, artificial look.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent chalk-like texture with realistic variations in line thickness.
- + Better preservation of the requested layout and menu items.
- + Captures the cozy café atmosphere with warm lighting on the frame.
- − Includes several spelling errors such as 'Risoktso', 'Mashroom', and 'Octpus'.
- − Repeats words like 'with with' and overlaps prices at the end of lines.
OmniGen v2
- + Text is highly legible even with smaller handwriting.
- + Follows the prompt for a clean wooden frame borders.
- + Correctly spells 'April 30, 2026' in a clear format.
- − Significant spelling errors throughout the menu items such as 'Specals', 'Lemont', and 'Musonhom'.
- − The text composition is cluttered and lacks the requested 'elegant cursive' for the title.
- − The handwriting looks more like a digital brush than authentic chalk.
Verdict: FLUX.1 Kontext [dev] is the winner because it captures the authentic aesthetics of a chalk menu, including the variations in handwriting style and texture requested in the prompt. While both models struggled with spelling the complex menu items, OmniGen v2 failed significantly more on word accuracy and overall layout balance.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Successfully followed the specific role-reversal instruction of the horse riding the astronaut.
- + High realistic rendering of textures on the spacesuit and horse fur.
- + Cinematic lighting with a clear sense of scale against the planet below.
- − The position of the horse is a bit ambiguous as it is behind the astronaut rather than clearly 'on top'.
- − The astronaut's hands are slightly distorted.
OmniGen v2
- + Bright, vibrant colors and clean lines.
- − Completely failed the negative/reversed prompt logic, showing an astronaut riding a horse.
- − The composition is generic and lacks the requested surrealism.
- − Visible anatomical issues where the astronaut's boots merge with the saddle/horse.
Verdict: FLUX.1 Kontext [dev] is the clear winner as it was the only model to attempt the role-reversal requested in the prompt ('horse on top, not vice versa'). OmniGen v2 ignored the specific instruction and generated a standard astronaut-on-horse image with several anatomical merging errors.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent adherence to the 'bored expression' of the passenger
- + Realistic integration of the capybara's anatomy into the driver's seat
- + High-quality textures and cinematic night lighting
- − One of the capybara's hands is not on the steering wheel, failing that specific prompt detail
OmniGen v2
- + Successfully places both paws on the steering wheel
- + Clean, vibrant colors in the taxi interior
- − The driver's hands are human rather than capybara paws
- − The passenger is in the front seat instead of the back seat as requested
- − The composition feels less like a real New York taxi interior
Verdict: FLUX.1 Kontext [dev] produced a much more atmospheric and realistic image that correctly placed the passenger in the back seat with the requested bored expression. While OmniGen v2 followed the instruction for both hands on the wheel, it mistakenly gave the capybara human hands and failed the spatial requirement of putting the passenger in the back.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent typography for the main title and date/location details
- + Clear implementation of the thorn border and twisted tree silhouettes
- + Very high resolution and clean digital illustration style
- − Text inside the scroll banner is garbled and unreadable
- − Lacks the requested 'dark parchment' texture, appearing more like a flat black digital poster
OmniGen v2
- + Beautiful parchment texture and moody night sky background with atmospheric lighting
- + Includes atmospheric elements like spiderwebs and a glowing moon effect behind the pumpkin
- + Main 'Halloween Party Invitation' text is elegant and well-rendered
- − Fails significantly on the sub-text, mixing keywords together in a confusing layout
- − Incorrect spelling of 'Arches' as 'Arcas' and 'Arnes'
- − Scroll banner text is illegible
Verdict: FLUX.1 Kontext [dev] produced a much more functional invitation with sharp, legible event details, though it missed the parchment texture and struggled with the small scroll banner. OmniGen v2 captured the 'vintage' and 'moody' aesthetic much better with its background and lighting, but the text at the bottom is messy, repetitive, and contains several spelling errors. FLUX.1 Kontext [dev] is the winner for providing a usable design despite the minor banner glitch.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent typography with clean, bold characters.
- + Perfectly captures the 'soft refined texture' and 'cartoon scene' aesthetic.
- + Very high clarity and clean composition.
- − The flag icon is stylized and doesn't represent a real flag accurately.
- − The sushi piece is a bit generic and simple in its construction.
OmniGen v2
- + Features a more complex and visually interesting diorama base.
- + Good adherence to the isometric perspective.
- + Text rendering is clean with nice shadow depth.
- − The flag icon is incorrect for Japan (red and yellow).
- − The sushi textures look a bit muddy and AI-generated compared to Model A's cleanliness.
- − The layout feels slightly cramped towards the center.
Verdict: FLUX.1 Kontext [dev] is the winner for its superior visual clarity and adherence to the requested 'soft refined textures' and 'ultra-clean' aesthetic. While OmniGen v2 provided a better diorama base, its textures were less polished and it failed to provide an accurate flag icon for Japan, whereas FLUX.1 Kontext [dev] felt more professional and intentional in its design.
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.1 Kontext [dev]
OmniGen v2
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent dynamic motion with animals actively pouncing and playing.
- + Superior photorealistic lighting and depth of field.
- + Natural fur textures and convincing environment integration.
- − Missed several requested animal types, showing multiple kittens/pups instead of fox and bunny.
- − The lighting is slightly hazy, which obscures some fine details.
OmniGen v2
- + Successfully included different species including the tabby kitten and fox-like kit.
- + Vibrant colors and clear god rays from the sunrise.
- + Captures the 'big expressive eyes' and 'wholesome vibe' very effectively.
- − Style is more '3D animation' or 'digital art' than the requested hyper-photorealism.
- − Static composition lacks the 'tumbling' and 'chasing' action requested.
- − The bunny is substituted or merged into a fox/cat hybrid appearance.
Verdict: FLUX.1 Kontext [dev] wins on pure technical quality and realism, capturing a believable sense of motion and light, though it failed to generate the specific variety of species requested. OmniGen v2 adhered better to the prompt's character list and color palette, but its output looks like a high-end cartoon rather than the requested masterpiece photograph.
Studio Ghibli Anime Style
EditingEdit instruction
“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”
Source
FLUX.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Successfully maintains the exact clothing patterns and colors from the original meme.
- + Captures the distinctive facial expressions, particularly the jealousy of the woman on the right.
- + Preserves the depth of field and street layout of the source image.
- − The art style leans more toward generic modern anime than the specific Ghibli aesthetic.
- − The lighting feels a bit flat compared to the requested 'warm, nostalgic mood'.
OmniGen v2
- + Excellent adherence to the Studio Ghibli art style with soft, rounded character designs.
- + Colors and lighting perfectly capture the warm, nostalgic, hand-painted feel requested.
- + Great attention to detail in the background, including the addition of greenery and softer street textures.
- − Fails to capture the jealous expression of the woman on the right, making everyone look happy.
- − Modifies the man's shirt pattern significantly compared to the original.
Verdict: FLUX.1 Kontext [dev] is much better at preserving the narrative and specific character details of the original meme, such as the plaid pattern and the expressions. However, OmniGen v2 far exceeds it in stylistic adherence, providing the soft, nostalgic, hand-painted aesthetic specific to Studio Ghibli, even though it loses the 'jealous' emotional context of the scene.
Golden Hour Stroll
Image EditingEdit instruction
“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
FLUX.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent preservation of the subject's face and original features
- + High resolution and consistent texture
- + Subtle but realistic wind effect on the hair
OmniGen v2
- + Successfully added more visible blowing leaves as requested
- + Stronger sense of motion in the hair flow
- − Noticeable changes to the subject's face compared to the source image
- − Leaves appear slightly artificial and disconnected from the depth of the scene
- − Loss of original detail in the background foliage
Verdict: FLUX.1 Kontext [dev] did an excellent job of preserving the identity of the person and the dog from the source image while adding subtle wind effects. OmniGen v2 added more 'dynamic' elements like orange leaves and windier hair, but it failed the preservation aspect by significantly altering the woman's facial features and changing the lighting and texture of the overall scene.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Perfect text rendering of the full name and date
- + Clean and bold minimalist vector style
- + Accurate representation of a cloche dome
- − Missed the request for a banner element
- − Typography is a bit more modern than 'classic'
OmniGen v2
- + Successfully included the banner and cloche as requested
- + Gently balanced composition with a vintage feel
- + Good use of the requested warm brown and cream tones
- − Significant spelling error in the main text ('CAFFFLORIN')
- − Poor preservation of 'Est.' which looks merged with other elements
Verdict: FLUX.1 Kontext [dev] produced a much higher quality logo by accurately rendering all text and maintaining a professional vector aesthetic, despite missing the requested banner. OmniGen v2 followed the composition instructions more closely (including the banner and cloche), but failed significantly on text accuracy by misspelling the primary brand name.
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.1 Kontext [dev]
OmniGen v2
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Successfully captures the requested navy and muted red color palette.
- + Includes a larger number of icon steps which better fits the sequential nature of the prompt.
- + Displays more legible, bold header text.
- − Severe text garbling and overlapping letters in the secondary descriptions.
- − The icons are messy and lack the 'clean' vector feel requested.
- − Misspells the main heading as 'APOLO' and 'TRANGULIITY'.
OmniGen v2
- + Excellent adherence to the clean, modern vector infographic aesthetic.
- + Very crisp icons and structured, professional layout.
- + Closer to the request for 'consistent iconography' with symmetrical and geometric shapes.
- − Incorrectly identifies the mission as 'Apolo 17' instead of Apollo 11.
- − Includes nonsensical text labels like 'NSA' and 'LANDEN ATE COLINS'.
- − Fails to reach the 'Landing' stage properly in terms of logical sequence.
Verdict: Both models struggled with the specific text requirements, but OmniGen v2 produced a much higher quality visual design that accurately reflects the 'modern vector infographic' style requested. While FLUX.1 Kontext [dev] attempted more steps of the mission, its messy execution and illegible typography make it less effective as a poster than the cleaner layout of OmniGen v2.
Next steps
Explore each model
Unified multimodal model for text-to-image generation, instruction-guided image editing, personalized generation, and virtual try-on