OpenAI's legacy image generation model supporting generations, edits with masks (inpainting), and variations
Settled by community votes across 3 shared challenges, with an AI judge weighing in on each.
DALL-E 2
17.7
arena score
#37 of 44 in Text-to-Image
Skill signature
Not enough comparable category data
The chart appears once both models have ratings across at least three shared arena categories.
FLUX.1 [schnell] FP8
17.8
arena score
#36 of 44 in Text-to-Image
Vote tally
Where the votes landed
DALL-E 2
0.0%
win rate
Ties
50.0%
FLUX.1 [schnell] FP8
50.0%
win rate
0.0%
50.0% ties
50.0%
Shared challenges
3
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
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.”
DALL-E 2
FLUX.1 [schnell] FP8
0% wins
0% ties
100% wins
AI Judge Analysis
DALL-E 2
- + The texture looks like genuine chalk strokes on a blackboard.
- − The text is completely illegible and does not follow the prompt.
- − The resolution is low and the image is heavily cropped.
- − It fails to render the specific menu items requested.
FLUX.1 [schnell] FP8
- + Successfully renders the specific date and most keywords from the menu items.
- + High visual quality with realistic lighting and a clear café setting.
- + The composition is well-balanced and resembles a professional chalkboard.
- − The handwriting looks more like a digital marker font than actual chalk on a board.
- − Includes several repetitive and garbled phrases instead of the clean list requested.
- − Significant spelling errors such as 'Risortto', 'Octemon', and 'Buthowith'.
Verdict: FLUX.1 [schnell] FP8 is the clear winner as it followed the prompt's instructions for specific text and date, whereas DALL-E 2 produced incomprehensible gibberish. While FLUX.1 struggled with perfect spelling and the 'chalk' texture feels a bit too much like a digital pen, it provided a coherent and aesthetically pleasing image that actually functions as a menu.
The Reversed Rodeo
Text-to-Image“Horse riding astronaut in space — horse on top, not vice versa. Surreal, highly detailed, cinematic.”
DALL-E 2
FLUX.1 [schnell] FP8
0% wins
100% ties
0% wins
AI Judge Analysis
DALL-E 2
- + Successfully captures a cinematic space background with realistic-looking stardust and astronomical bodies.
- + Correctly places an astronaut and a horse in the scene together.
- − Completely failed the negative constraint to have the horse on top of the astronaut.
- − Poor image quality with significant artifacts, smudging, and a low-resolution appearance.
- − Anatomy of both the horse and astronaut is distorted and lacks detail.
FLUX.1 [schnell] FP8
- + Followed the difficult logical constraint of placing a horse on top of an 'astronaut' (interpreted here as astronaut equipment).
- + Exceptional visual clarity and high-resolution details on the fur and mechanical parts.
- + Dynamic composition with a surreal, cinematic lighting style.
- − The astronaut is represented as robotic equipment/backpack rather than a human in a suit.
- − Anatomical strangeness where a second horse head/neck appears to emerge from the back.
Verdict: DALL-E 2 completely failed the core logical reversal requested in the prompt, providing a standard, low-quality image of an astronaut riding a horse. FLUX.1 [schnell] FP8 successfully interpreted the surreal 'horse on top' instruction with high fidelity and cinematic detail, though it translated the 'astronaut' into mechanical equipment.
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.”
DALL-E 2
FLUX.1 [schnell] FP8
AI Judge Analysis
DALL-E 2
- + Features a hand-drawn, authentic vintage aesthetic
- + Includes the requested twisted trees and bats correctly
- − Text is completely illegible and nonsensical
- − Low resolution and lacks the requested central glowing jack-o-lantern
- − Missing specific event details like date and location
FLUX.1 [schnell] FP8
- + Excellent composition with a clear central glowing jack-o-lantern
- + High resolution with clean, cinematic lighting
- + Successfully includes the specific date, time, and location requested in the prompt
- − Notable typos in the sub-text and labels such as 'Tine' and 'Theaches'
- − The border is slightly simplistic despite the spooky theme
Verdict: FLUX.1 [schnell] FP8 is the clear winner as it followed every instruction, including the layout of a central pumpkin and specific event details, while maintaining high visual quality. DALL-E 2 failed to include the primary subject (the pumpkin) and produced entirely garbled text that makes the invitation unusable.
Next steps
Explore each model
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