Unified multimodal model for text-to-image generation, instruction-guided image editing, personalized generation, and virtual try-on
Settled by community votes across 13 shared challenges, with an AI judge weighing in on each.
OmniGen v2
16.8
arena score
#57 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Stable Diffusion 3.5 Large
22.8
arena score
#30 of 62 in Text-to-Image
Vote tally
Where the votes landed
OmniGen v2
0%
win rate
Ties
0%
Stable Diffusion 3.5 Large
0%
win rate
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Excellent adherence to spatial relationships defined in the prompt
- + Clean and high-quality rendering of glass and soft lighting
- + Good composition with all elements clearly visible
Stable Diffusion 3.5 Large
- + Realistic glass textures and reflections
- + Accurate lighting direction from the left
- − Failed the spatial request by putting the book inside the cube and the sphere on the book
- − The plant is mostly above rather than behind the cube
- − Distracting background elements not requested in the prompt
Verdict: OmniGen v2 followed all spatial instructions perfectly, placing the sphere inside the cube and the book on top. Stable Diffusion 3.5 Large struggled with the complex prepositional logic, reversing the positions of the sphere and the book and placing the book inside the glass container instead.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Excellent reflection quality on the wet pavement
- + Good adherence to the 50mm shallow depth of field request
- + The man's clothing and facial features are well-rendered
- − The man is just standing with the bike rather than actively 'repairing' it
- − The image looks slightly artificial or over-processed, missing the 'no stylization' goal
- − The bike geometry is slightly warped near the pedals
Stable Diffusion 3.5 Large
- + Stronger adherence to the 'repairing' action with a more candid pose
- + Captures the requested 'imperfect framing' and cinematic realism effectively
- + More natural skin texture and grittier, realistic atmosphere
- − Anatomy issues with the man's hands appearing mangled
- − The bicycle frame has some structural inconsistencies
- − Background car lacks the 'motion blur' requested, appearing mostly static
Verdict: Stable Diffusion 3.5 Large wins because it captures the narrative of the prompt much more effectively, showing an elderly man actually engaged in repairing a bike with a gritty, realistic aesthetic. While OmniGen v2 has cleaner technical rendering and better reflections, it feels like a staged studio photo rather than the candid street photography requested.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Excellent depiction of warm torchlight reflecting on metal
- + Accurate inclusion of beads in the braided hair
- + Very clean face and eye rendering
- − Character looks too clean and youthful for the 'battle-worn' descriptor
- − The dirt on the skin looks like artificial spots rather than grime
Stable Diffusion 3.5 Large
- + Perfectly captures the 'battle-worn' aesthetic with realistic scars and dirt
- + Exceptional detail on the engraved plate armor and chainmail underlayer
- + Intense, lifelike gaze that fits the paladin archetype
- − Missed the request for beads in the hair braids
- − The light is more like overcast daylight than warm torchlight
Verdict: Stable Diffusion 3.5 Large is the clear winner for its superior texture and adherence to the 'battle-worn' and 'highly detailed' aspects of the prompt, creating a much more believable character. While OmniGen v2 correctly included the beads and better torchlight coloring, the overall image feels too much like a modern person in a costume compared to the grit and detail of Stable Diffusion 3.5 Large.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Stronger adherence to the multi-column menu layout
- + Appealing use of color block accents for a vibrant look
- + Effective integration of food photography within the menu sections
- − Numerous spelling errors in headings like 'RESTAURATED MENTS' and 'PIZZZZAN'
- − The font style leans more towards a distressed or retro slab-serif rather than a clean modern sans-serif
Stable Diffusion 3.5 Large
- + Excellent typography with a bold, clean sans-serif focus centered on the page
- + High-quality, vibrant food photography in a clear grid pattern
- + Superior text rendering for the primary 'Menu' heading
- − Text content in the sections is largely gibberish
- − The grid layout feels slightly busy and crowds the central text area
Verdict: Stable Diffusion 3.5 Large produced a much more professional-looking design with superior typography and high-quality food photography that felt modern and clean. While OmniGen v2 attempted a more complex layout, its severe spelling errors and less polished font choices made it feel less cohesive as a 'professional' 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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Successfully integrated all requested text with clear readability.
- + Includes the starburst element for the price as requested.
- + Clean graphic design style suitable for a quick-service advertisement.
- − Failed to create an 'exploded' view where components are suspended in mid-air.
- − The burger looks like a static illustration rather than having photorealistic detail.
- − The 'fiery' effect is more of a graphic halo than a realistic environment.
Stable Diffusion 3.5 Large
- + Excellent photorealistic details on the meat, cheese, and bun.
- + Features high-quality textures for the fire and glowing embers background.
- + Better atmospheric lighting and a much more dynamic feel.
- − Completely failed to include any of the requested text.
- − Failed to produce the 'exploded' view; the burger is fully assembled.
- − Missing the required starburst element.
Verdict: OmniGen v2 successfully followed the complex text instructions and formatting but failed to deliver the specific 'exploded' composition or photorealistic quality. Stable Diffusion 3.5 Large produced a much more visually impressive and realistic image with great atmospheric detail, but it failed to render any of the requested text or the specific deconstructed burger layout. OmniGen v2 is the winner for functional prompt adherence, whereas Stable Diffusion 3.5 Large is visually superior but functionally incomplete.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + The text has a convincing chalk texture and handwriting style.
- + The formatting of the chalkboard is clear and legible.
- − Contained several spelling errors like 'SPECALS' and 'lemont'.
- − The text arrangement becomes jumbled and illogical in the middle section.
Stable Diffusion 3.5 Large
- + Excellent environmental composition and 'cozy café' atmosphere.
- + The chalkboard layout is visually appealing and consistent with a real menu.
- − Failed the specific date request, using '2024' instead of '2026'.
- − The text quickly dissolves into 'gibberish' for all the specific menu items.
Verdict: OmniGen v2 does a superior job with the specific handwriting texture and follows the text prompt more closely, despite some spelling errors. Stable Diffusion 3.5 Large creates a much more beautiful and atmospheric image of a café, but it fails significantly on text legibility and accuracy, missing the requested date and scrambling the menu items.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Clean, vector-like aesthetic with high contrast
- + Correctly places the astronaut in a space suit
- + Clearly defined subjects with no blurring or noise artifacts
- − Failed the negative constraint; the astronaut is riding the horse instead of the horse riding the astronaut
- − Lack of cinematic depth; feels more like a 2D illustration
- − Oversimplified space background
Stable Diffusion 3.5 Large
- + Highly cinematic atmosphere with excellent light and cloud particles
- + Richly detailed astronaut suit and horse texture
- + Dynamic composition with a sense of movement
- − Failed the specific negative constraint; horse is being ridden by the astronaut
- − Anatomical issues with the horse's legs and hooves
- − Face of the horse is slightly distorted
Verdict: Both models failed the specific logic constraint of the prompt ('horse on top, not vice versa'), defaulting to the common trope of an astronaut riding a horse. Stable Diffusion 3.5 Large is the superior image due to its cinematic quality, atmospheric depth, and intricate detailing, whereas OmniGen v2 produced a flat, simplistic illustration with less visual interest.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Excellent texture on the capybara's fur
- + Correct positioning of the businesswoman in the back seat
- + Clean and photorealistic lighting
- − Human hands are emerging from the capybara's sleeves instead of paws
- − Businesswoman is holding the steering wheel from the back seat
- − Hybrid human-animal anatomy is disconcerting
Stable Diffusion 3.5 Large
- + Correctly depicts capybara paws interacting with the steering wheel
- + Very detailed and complex clothing textures
- + Dynamic cinematic lighting with good bokeh
- − Completely failed to include the passenger in the back seat
- − The capybara's body and legs are more human-proportioned than requested
- − Composition is a bit cramped
Verdict: OmniGen v2 correctly included both the capybara driver and the human passenger as requested, but failed significantly on anatomy by giving the capybara human hands and having the passenger also steer the car from the back. Stable Diffusion 3.5 Large followed the capybara details more accurately (paws on wheel), but completely missed the requirement for a passenger in the back seat. Stable Diffusion 3.5 Large is the better image in terms of detail and logic, despite the missing character.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Excellent typography rendering for the main title and specific event details.
- + Clean composition that balances text and illustration effectively.
- + Accurate adherence to the square format requested.
- − The 'scroll banner' is illegible and garbled compared to the other text.
- − The overall aesthetic feels a bit more like a modern vector graphic than a 'vintage gothic parchment'.
Stable Diffusion 3.5 Large
- + Superb vintage gothic aesthetic with highly detailed parchment textures and hand-drawn style illustration.
- + Correct rendering of the small scroll banner text.
- + Excellent moody, cinematic lighting and atmosphere.
- − Completely failed to include the specific event details (Date, Time, Location) at the bottom.
- − The main title text is simple and lacks the 'elegant gothic' flair requested.
Verdict: OmniGen v2 wins on functional utility because it successfully included almost all the requested text, including the specific date and location, whereas Stable Diffusion 3.5 Large ignored them entirely. However, Stable Diffusion 3.5 Large produced a much more visually stunning and stylistically accurate 'vintage gothic' illustration. If the specific event details were not a priority, Stable Diffusion 3.5 Large would be the artistic winner, but for a functional invitation, OmniGen v2 is preferred.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Excellent typography rendering with bold 3D effects.
- + Strict adherence to the light blue background and minimalist diorama base requested.
- + Perfectly centered and clean composition.
- − The flag icon is incorrect, showing yellow and red stripes instead of the Japanese flag.
- − The sushi pieces contain odd black squares in the rice that look like glitches.
Stable Diffusion 3.5 Large
- + High level of detail in textures, particularly the sushi rice and fish.
- + Includes a correct Japanese flag icon.
- + Strong sense of 3D miniature scale with varied sushi types.
- − Failed to place the text at the 'top-center', instead placing it on a sign.
- − Does not use a solid light blue background, including a textured floor/surface instead.
- − The composition is busier than requested, ignoring the 'minimal' garnish instruction.
Verdict: OmniGen v2 followed the layout and background instructions much better, producing a clean graphic design feel with clear central text, though it failed on the flag accuracy. Stable Diffusion 3.5 Large produced much higher quality textures and a correct flag, but it ignored the specific text placement and background simplicity in favor of a more complex scene. OmniGen v2 is preferred for professional graphic layout, whereas Stable Diffusion 3.5 Large is better for realistic texture work.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Strong adherence to the morning sunrise lighting and god rays
- + Coherent animal groupings with high contrast
- − Failed to provide all four requested animals, missing the bunny
- − The style is very cartoony/illustrative rather than 'hyper-photorealistic'
- − Static composition that doesn't capture the 'playfully chasing' or 'tumbling' action
Stable Diffusion 3.5 Large
- + Includes all four requested animals: puppy, kitten, bunny, and fox kit
- + Achieved a much more photorealistic look with believable fur textures
- + Dynamic composition capturing movement and playfulness as requested
- − The fox and kitten look quite similar in facial structure
- − Lighting is a bit hazy, losing some of the 'sparkle' detail in the background
Verdict: Stable Diffusion 3.5 Large followed the prompt much more accurately, including all four specified animals and successfully capturing the 'hyper-photorealistic' and 'playful' aspects. OmniGen v2 failed on the animal count and produced a highly stylized, almost Pixar-like cartoon image that ignored the request for realism.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Strong minimalist vector aesthetic
- + Excellent symmetry and clean line work
- + Clear and legible 'Est. 1720' text
- − Spelling error in the main text ('CAFFFLORIN')
- − Steam effect is overly simplistic compared to the rest of the logo
Stable Diffusion 3.5 Large
- + Correctly spelled 'Caffè Florian'
- + Excellent vintage texture on the background
- + Includes more intricate graphic ornaments
- − Slight misspelling of 'Caffé' (extra accent/letter e overlap)
- − Graphic composition is a bit cluttered compared to a minimalist style
Verdict: Stable Diffusion 3.5 Large is the winner as it correctly rendered the full restaurant name and captured the requested 'subtle texture' significantly better than OmniGen v2. While OmniGen v2 had a cleaner minimalist aesthetic, the major spelling error in the primary brand name makes it less usable.
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.”
OmniGen v2
Stable Diffusion 3.5 Large
AI Judge Analysis
OmniGen v2
- + Clean vector layout that adheres well to the 'infographic poster' format.
- + Successfully used the requested NASA-inspired color palette.
- + Includes distinct icons for different stages of the mission.
- − Confused Apollo 11 with Apollo 17 and NASA with NSA.
- − Text is mostly gibberish or misspelled.
- − The icons do not accurately represent the specific mechanical steps requested (e.g., Saturn V icon).
Stable Diffusion 3.5 Large
- + Rich in detail with a sophisticated technical layout.
- + Captures the NASA aesthetic and scale effectively.
- + Good use of the color palette and background contrast.
- − Inaccurate rocket design (Space Shuttle hybrid instead of Saturn V).
- − Fails to follow the 'steps' structure or the specific requested icons.
- − The composition is cluttered with illegible micro-text.
Verdict: OmniGen v2 better understands the requested 'infographic' format and provides a cleaner, more organized layout that mimics a real poster, despite significant spelling errors and getting the mission number wrong. Stable Diffusion 3.5 Large produces a more visually striking illustration, but it fails to follow the logical progression of steps or the specific flat-vector icon requirements. OmniGen v2 is the winner for better prompt adherence regarding style and layout.
Next steps
Explore each model
Stability AI's 8.1-billion parameter Multimodal Diffusion Transformer (MMDiT) text-to-image model featuring improved image quality, typography, complex prompt understanding, and resource-efficiency