FP8 quantized variant of Black Forest Labs' FLUX.1 [schnell] model, offering ~2x faster inference with reduced precision while maintaining high-quality image generation in 4 steps
Settled by community votes across 6 shared challenges, with an AI judge weighing in on each.
FLUX.1 [schnell] FP8
19.0
arena score
#47 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Imagen 3.0 Generate 002
21.0
arena score
#36 of 62 in Text-to-Image
Vote tally
Where the votes landed
FLUX.1 [schnell] FP8
0%
win rate
Ties
0%
Imagen 3.0 Generate 002
0%
win rate
Shared challenges
6
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
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 [schnell] FP8
Imagen 3.0 Generate 002
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Very clean layout with clear section headers
- + Professional white background follows minimalist prompt strictly
- + Excellent color vibrancy in food photography
- − Nonsense words like 'ORCETERS' and 'SECCER'
- − Food photography lacks high-resolution detail on close-up inspection
- − The menu header is clipped and visually awkward
Imagen 3.0 Generate 002
- + Strong grid-based composition creates a consistent visual pattern
- + High-quality, realistic food photography
- + Better integration of text sections within the grid layout
- − Sections are repetitive (e.g., 'MAINS' and 'PIZZA' appear multiple times)
- − The background is a textured wall rather than a clean, flat minimalist white background
- − Typography is somewhat cluttered for a minimalist design
Verdict: FLUX.1 [schnell] FP8 offers a more traditional and functional menu layout that adheres well to the minimalist requirement, though it suffers from poor text legibility and nonsensical categories. Imagen 3.0 Generate 002 produces much higher quality food imagery and a more interesting grid, but fails the professional menu structure by repeating the same category names multiple times in a confusing manner. FLUX.1 [schnell] FP8 is the preferred choice for a layout closer to a real-world menu design.
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 [schnell] FP8
Imagen 3.0 Generate 002
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Strong sense of depth with the foreground flames and background smoke
- + Dynamic scattering of small ingredients adds to the motion effect
- − Severe spelling errors including 'LIIMITED' and 'NEEY'
- − Failed to follow the price instruction, displaying '€69' instead of '€6.99'
- − The burger is mostly intact rather than 'exploded' with suspended components
Imagen 3.0 Generate 002
- + Perfectly executed the 'exploded' burger concept with clearly suspended layers
- + Accurate text rendering for the primary title and the specific price
- + Excellent lighting integration between the fiery background and the burger elements
- − Gibberish text above the price in the starburst
- − Composition is a bit vertically cramped at the bottom
Verdict: Imagen 3.0 Generate 002 is the clear winner as it followed all complex instructions, including the specific exploded layout of the burger and accurate text rendering. FLUX.1 [schnell] FP8 failed significantly on the text, producing multiple typos and the wrong price, and it did not properly separate the burger components as requested.
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 [schnell] FP8
Imagen 3.0 Generate 002
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent chalk-like texture on the lettering
- + Realistic framing and lighting
- − Numerous spelling errors like 'Risortto', 'Griffled', and 'Octemon'
- − Fails to render the text in cursive as requested
- − Repetitive lines with conflicting prices
Imagen 3.0 Generate 002
- + Successfully captures an elegant cursive-leaning handwriting style
- + Follows price prompts more accurately ($24, $28, $9)
- + Better overall composition and layout on the board
- − Includes some spelling gibberish like 'Octpsuin' and 'Choouip'
- − Handwriting looks slightly more like a digital brush than physical chalk compared to model A
Verdict: Imagen 3.0 Generate 002 is the clear winner as it followed the stylistic instruction for elegant cursive handwriting and adhered closer to the requested prices. While both models struggled with spelling complex words, FLUX.1 [schnell] FP8 produced redundant lines and failed to incorporate the cursive requirement.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
FLUX.1 [schnell] FP8
Imagen 3.0 Generate 002
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Attempts the difficult spatial instruction of having the horse on top.
- + Creates a very surreal and dreamlike composition with the Earth backdrop.
- + The lighting and textures on the horse are cinematically rendered.
- − Anatomy is a mess, appearing as a two-headed horse or multiple horses fused into mechanical parts.
- − Lacks a clear human astronaut figure, replacing it with a boxy satellite-like object.
Imagen 3.0 Generate 002
- + Excellent visual clarity and beautiful celestial background.
- + Very clean anatomy and realistic rendering of the horse and spacesuit.
- + High level of detail in the textures of the horse's fur and the suit material.
- − Completely failed the negative/inverted prompt instruction by placing the astronaut on top.
- − Lacks the 'surreal' quality requested, opting for a standard sci-fi interpretation.
Verdict: This was a highly specific prompt testing spatial reasoning. FLUX.1 [schnell] FP8 correctly attempted the 'horse on top' instruction but failed to render a coherent subject, resulting in a confusing anatomical mess. Imagen 3.0 produced a much more beautiful and polished image, but it failed the primary core instruction by placing the astronaut on the horse in the traditional manner.
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 [schnell] FP8
Imagen 3.0 Generate 002
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent 3D rendering with soft, refined textures
- + Clean and modern isometric composition
- + Good adherence to the light blue background and diorama base requests
- − Failed the text prompt, repeating Japan twice and misspelling it with a red dot
- − Composition is a bit crowded within the small tray
Imagen 3.0 Generate 002
- + Perfect adherence to text and flag instructions
- + Highly accurate 45-degree isometric projection
- + Excellent variety of sushi types following the 3D cartoon style
- − The lighting is a bit flat compared to the requested refined PBR textures
- − The text is placed in the top-left rather than top-center as requested
Verdict: Imagen 3.0 Generate 002 follows the specific prompt instructions significantly better, accurately rendering both the 'JAPAN' and 'SUSHI' text along with a flag icon. While FLUX.1 [schnell] FP8 has slightly more sophisticated lighting and texture work, its failure to render the requested text correctly makes it less successful overall.
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 [schnell] FP8
Imagen 3.0 Generate 002
AI Judge Analysis
FLUX.1 [schnell] FP8
- + Excellent vibrant lighting and butterflies
- + High level of detail in the fur texture
- + Captures a joyful, expressive mood
- − Failed to include a rabbit
- − Includes five animals instead of the requested four
- − Anatomical issues with the animals on the right merging together
Imagen 3.0 Generate 002
- + Perfect adherence to the prompt, including the puppy, kitten, bunny, and fox
- + Natural, photorealistic fur and lighting
- + Realistic composition with a clear 'tumbling' action from the bunny
- − Dew sparkles look slightly digital like white dots
- − Lower number of butterflies compared to Image A
Verdict: Imagen 3.0 Generate 002 is the clear winner as it successfully included all four requested animals, including the baby bunny which FLUX.1 [schnell] FP8 missed entirely. While FLUX.1 produced more vibrant colors, it failed on count and species accuracy, whereas Imagen 3.0 provided a more realistic and faithful interpretation of the prompt.
Next steps
Explore each model
Google's Imagen 3.0 text-to-image generation model, producing high-quality images with improved detail and lighting