How 21 AI image models render: “A confident adult Korean witch with cat-eye glasses and orange hair flies on her broom. She wears a fitted short black dress and chunky shoes. A carved pumpkin also rides the broomstick. Moon, distant city. The witch is smiling. Stylized semi-realistic digital character painting, expressive animation-inspired facial proportions, large eyes and strong brows, simplified natural anatomy with convincing three-dimensional volume. Painterly shape-based rendering, minimal linework, soft blended facial shading contrasted with broad angular brushstrokes on clothing. Warm luminous skin with peach and terracotta undertones, subtle rosy cheeks and nose, restrained highlights. Hair rendered as flowing sculpted masses with broad highlights and a few fine flyaway strands. Muted charcoal and earthy neutral palette, soft directional studio lighting, carefully rendered eyes and lips, selectively loose edges and unfinished brushwork.”

Same prompt, sent verbatim, at each model's default settings and a fixed seed. Judge: line quality, colour, style consistency. 166 community votes so far.

Community prompt by @dell · voting closed 2026-11-06

21 text-to-image models rendered this illustration & anime prompt with the identical text at their default settings and a fixed seed. GPT Image 2.5 Flare is the community favourite so far, winning 75% of its 16 head-to-head votes on this prompt. FLUX 3 Image Klein was the fastest at 6.2s.

What to look for: confident line work, consistent style from edge to edge, deliberate colour and readable shapes.

FAQ

Which AI model renders this prompt best?

GPT Image 2.5 Flare leads this prompt in the BudgetPixel AI Image Arena, winning 75% of its 16 blind head-to-head votes. Vote on the arena page to add your judgement.

Were the prompts adjusted per model?

No. Every model received the exact same prompt, unmodified, at its default settings, with no negative prompt or style suffix. Some models rewrite prompts internally, and that is part of what is being compared.

Can I run this prompt myself?

Yes. Each tile’s “Try” link opens the BudgetPixel workshop with this prompt and that model preselected.