Pick a model for each pane here, in one place — then hit Compare blind. Panes shuffle and hide their labels until you vote, so picking never spoils the comparison.
Pane A
Pane B
Opus 0·Sonnet 0
community votes
What is this?
A marketing website for a made-up product: an app that turns your travel photos into printed hardcover books. Both models wrote all the text and designed everything from scratch — no images or fonts allowed, so every visual is hand-drawn code. Judge it like a customer: does this look like a real company you would give your credit card to?
Generation stats
Fable 5 · low
~15,302 output tokens · $0.77 est. · 15m 59s
$50/1M output tokens
community votes: 0
Fable 5 · medium
~15,678 output tokens · $0.78 est. · 11m 39s
$50/1M output tokens
community votes: 0
Fable 5 · high
~21,072 output tokens · $1.05 est. · 46m 21s
$50/1M output tokens
community votes: 0
Fable 5 · xhigh
~17,117 output tokens · $0.86 est. · 40m 33s
$50/1M output tokens
community votes: 0
Fable 5 · max (post-ban)
~20,502 output tokens · $1.03 est. · 28m 24s
$50/1M output tokens
community votes: 0
Fable 5 (pre-ban)
~22,069 output tokens · $1.10 est.
$50/1M output tokens
community votes: 0
Opus 4.8 · low
~15,989 output tokens · $0.40 est.
$25/1M output tokens
community votes: 0
Opus 4.8 · medium
~16,254 output tokens · $0.41 est.
$25/1M output tokens
community votes: 0
Opus 4.8 · high
~18,884 output tokens · $0.47 est.
$25/1M output tokens
community votes: 0
Opus 4.8 · xhigh
~17,740 output tokens · $0.44 est.
$25/1M output tokens
community votes: 0
Opus 4.8 · max
~16,736 output tokens · $0.42 est.
$25/1M output tokens
community votes: 0
Opus 4.8 · ultracode
n/a* output tokens · — est.
$25/1M output tokens
community votes: 0
Sonnet 4.6 · low
~14,662 output tokens · $0.22 est.
$15/1M output tokens
community votes: 0
Sonnet 4.6 · medium
~16,260 output tokens · $0.24 est.
$15/1M output tokens
community votes: 0
Sonnet 4.6 · high
~17,136 output tokens · $0.26 est.
$15/1M output tokens
community votes: 0
Sonnet 4.6 · max
~19,302 output tokens · $0.29 est.
$15/1M output tokens
community votes: 0
Sonnet 5 · low
~14,067 output tokens · $0.21 est.
$15/1M output tokens
community votes: 0
Sonnet 5 · medium
~15,700 output tokens · $0.24 est.
$15/1M output tokens
community votes: 0
Sonnet 5 · high
~20,723 output tokens · $0.31 est.
$15/1M output tokens
community votes: 0
Sonnet 5 · xhigh
~19,221 output tokens · $0.29 est.
$15/1M output tokens
community votes: 0
Sonnet 5 · max
~14,577 output tokens · $0.22 est.
$15/1M output tokens
community votes: 0
Sonnet 5 · ultracode
n/a* output tokens · — est.
$15/1M output tokens
community votes: 0
Haiku 4.5
~7,947 output tokens · $0.04 est.
$5/1M output tokens
community votes: 0
GLM-5.2
~13,496 output tokens · — est.
per-1M output pricing not published
community votes: 0
GLM-5.2 · flash
~13,486 output tokens · — est.
per-1M output pricing not published
community votes: 0
GPT-5.5
~8,910 output tokens · $0.09 est.
$10/1M output tokens
community votes: 0
Gemini 3 Flash
~7,576 output tokens · $0.02 est.
$3/1M output tokens
community votes: 0
Kimi K2.7 Code
~12,783 output tokens · $0.05 est.
$4/1M output tokens
community votes: 0
Qwen3.7 Plus
~10,456 output tokens · $0.02 est.
$1.6/1M output tokens
community votes: 0
DeepSeek V4 Flash
~16,397 output tokens · $0.00 est.
$0.28/1M output tokens
community votes: 0
MiniMax M3
~2,721 output tokens · $0.00 est.
$1.2/1M output tokens
community votes: 0
Kimi K2.6
~16,630 output tokens · $0.07 est.
$4/1M output tokens
community votes: 0
Qwen3.7 Max
~10,748 output tokens · $0.02 est.
$1.6/1M output tokens
community votes: 0
DeepSeek V4 Pro
~16,422 output tokens · $0.00 est.
$0.28/1M output tokens
community votes: 0
MiniMax M2.7
~18,430 output tokens · $0.02 est.
$1.2/1M output tokens
community votes: 0
GPT-5.5 · low
~8,272 output tokens · $0.08 est.
$10/1M output tokens
community votes: 0
GPT-5.5 · medium
~10,578 output tokens · $0.11 est.
$10/1M output tokens
community votes: 0
GPT-5.5 · high
~11,199 output tokens · $0.11 est.
$10/1M output tokens
community votes: 0
GPT-5.5 · xhigh
~12,940 output tokens · $0.13 est.
$10/1M output tokens
community votes: 0
Mistral Large 2512
~12,429 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Mistral Medium 3.5
~7,018 output tokens · $0.01 est.
$2/1M output tokens
community votes: 0
Mistral Medium 3 (2508)
~15,458 output tokens · $0.03 est.
$2/1M output tokens
community votes: 0
Mistral Medium (2505)
~10,317 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Mistral Small 2603
~10,329 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Mistral Small 2506
~8,717 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Codestral 2508
~8,088 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Devstral 2512
~9,572 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Magistral Medium · reasoning
~5,345 output tokens · $0.01 est.
$2/1M output tokens
community votes: 0
Magistral Medium · no-reason
~7,383 output tokens · $0.01 est.
$2/1M output tokens
community votes: 0
Magistral Small · reasoning
~8,652 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Magistral Small · no-reason
~5,693 output tokens · $0.01 est.
$2/1M output tokens
community votes: 0
Ministral 14B
~9,830 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Ministral 8B
~8,749 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Ministral 3B
~9,818 output tokens · $0.02 est.
$2/1M output tokens
community votes: 0
Mistral Nemo
~2,556 output tokens · $0.01 est.
$2/1M output tokens
community votes: 0
Output-token counts are estimated from each entry's file size (chars ÷ 4), not a transcript, so they are approximate (shown with ~). Cost = estimated output tokens × the model's published output price (Fable 5 $50, Opus 4.8 $25, Sonnet 4.6 $15, Haiku 4.5 $5 per 1M; GLM-5.2 unpublished → —). Input tokens and reasoning/thinking overhead aren't counted, so true cost is higher — especially at higher effort levels. Where a third number appears it is wall-clock generation time; only runs generated after 2026-07-02 recorded it.
Exact prompt both models received
Show full prompt
You are producing one entry for a creative coding benchmark. Work fully autonomously - do not ask questions.
Deliverable: ONE single fully self-contained HTML file. All CSS and JS inline. Zero external resources: no CDNs, no web fonts, no external images, no network requests. All visuals via CSS, inline SVG, or canvas. Must work offline served from a local static server.
Quality bar: this entry will be judged side-by-side against another model's attempt at the exact same brief. Visual polish, originality, attention to detail, and flawless functionality all count. One shot - make it your best work.
TASK: Landing page for "Driftwood" - a fictional service that turns your phone's travel photos into beautifully printed hardcover story books using AI curation. Required sections: sticky nav, hero with primary CTA, social proof strip, 3-6 feature highlights, how-it-works (3 steps), pricing with 3 tiers (one highlighted), testimonials, FAQ (accordion), final CTA, footer. Fully responsive, smooth scroll, tasteful animations and micro-interactions, coherent distinctive art direction (not a generic Bootstrap look). Write all copy yourself in English - make it persuasive and human.