OpenAI's previous generation image model with higher quality than DALL-E 2 and support for larger resolutions
Settled by community votes across 1 shared challenge, with an AI judge weighing in on each.
DALL-E 3
18.5
arena score
#35 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.2 [klein] 4B
23.8
arena score
#22 of 44 in Text-to-Image
Vote tally
Where the votes landed
DALL-E 3
0%
win rate
Ties
0%
FLUX.2 [klein] 4B
0%
win rate
Shared challenges
1
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
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 3
FLUX.2 [klein] 4B
AI Judge Analysis
DALL-E 3
- + Exquisite visual detail with highly intricate gothic borders and textures
- + Excellent captures of the requested vintage parchment feel and cinematic lighting
- + Creative interpretation of the central jack-o-lantern within a carved frame
- − Failed significantly on text rendering, with most words being garbled gibberish
- − The 'Halloween Party Invitation' text is split and visually cluttered
FLUX.2 [klein] 4B
- + Successfully included all layout elements including the scroll banner and event details at the bottom
- + Much better legibility of the core event details like the date and location
- + Strong adherence to the 'thorns' element within the webbed border
- − Significant spelling errors in the main title ('Hallbwom Party niisation')
- − Lettering in the scroll contains a typo ('friglts' instead of 'frights')
Verdict: DALL-E 3 creates a much more atmospheric and visually stunning gothic artwork, but it fails completely at the requested text transcription. FLUX.2 [klein] 4B follows the layout and text instructions much more accurately, and although it contains spelling errors, the information is legible enough to be functional as an invitation.
Next steps
Explore each model
Black Forest Labs' compact, open-source image generation model with sub-second inference, optimized for production and near real-time applications with multi-reference support