TASK 01 / HTML · V1 · Website

Aevum Luxury Watch Landing Page

Aevum 奢侈腕表落地页

4 runs · 01.A.01 · HTML

Compare works →
TASK 01 · 01.A.01 · HTMLAevum Luxury Watch Landing Page
Open independent copy ↗
Aevum Luxury Watch Landing Page

Preview note: Embedded storage and downloads can be restricted by browser sandboxing; use the standalone view for full behavior.

ORIGINAL PROMPT

Original prompt

PROMPT/Phase1_TEST_PROMPTS_BILINGUAL.md · The original prompt is kept verbatim

Source ↗
ORIGINAL PROMPT

Original prompt

Build a beautiful landing page for a fictional luxury watch brand called "Aevum".
Use only HTML, CSS and JavaScript.
Do not use external images or libraries.
Make it feel like a real premium product website.
TRANSLATION

Chinese translation

为名为“Aevum”的虚构奢侈腕表品牌制作一个精美的落地页。只使用 HTML、CSS 和 JavaScript,不使用外部图片或第三方库,并使页面具备真实高端产品官网的质感。

Expected outcome

The deliverable should be a complete landing page that opens directly in a modern browser. It should establish a coherent premium visual language through branding, a strong hero, product storytelling, collections or key selling points, craftsmanship and material details, clear calls to action, and a finished footer. All visuals should be produced with HTML, CSS, inline SVG, Canvas, or procedural graphics, with no external imagery, fonts, frameworks, or component libraries. The result should combine narrative, hierarchy, responsive behavior, interaction feedback, and accessibility so it feels publishable rather than like a basic prototype.

Additional verification guidance

  • Open the page offline and confirm there are no external requests, missing assets, or console errors; verify that only HTML/CSS/JavaScript is used and all imagery is local or generated inline.
  • Test desktop, tablet, and mobile viewports for navigation, hero, content, forms, and CTAs, with no overflow, clipping, overlap, or unreadable text.
  • Review brand consistency, hierarchy, typography, spacing, color, restrained motion, and premium feel; verify correct feedback for buttons, menus, configurators, or forms.
  • Complete primary flows by keyboard and check semantic headings, form labels, visible focus, sufficient contrast, and reduced-motion support.
THREE INTERPRETATIONS

Same task, three model interpretations

All three results use the same width and the same preview ratio; no model is marked as best.

COMPARISON METRICS

Comparison metrics

Cross-model comparison must note differences in harness, budget and isolation.

MetricDeepSeek-V4.1-Flashrun-deepseek-v4-1-flash-exp-0910-task-01-r1Muse Spark 1.3run-muse-spark-1-3-task-01-r1GPT-5.6 Solrun-gpt-5-6-sol-task-01-r1K3run-k3-task-01-r1
Runtime4:512:136:0633:08
API calls2410421
Tool calls229322
Failures / retries4001
Token throughput2M234.2K134.5K1.1M
AI review96/10094/10090/10098/100
RUN RECORDS

Run records

Two runs of the same task version can be compared side by side.

01.A.01 · run-deepseek-v4-1-flash-exp-0910-task-01-r1DeepSeek-V4.1-Flash● Archived
Runtime
4:51
Token throughput
2M
API calls
24
View archive record →
01.B.01 · run-muse-spark-1-3-task-01-r1Muse Spark 1.3● Archived
Runtime
2:13
Token throughput
234.2K
API calls
10
View archive record →
01.C.01 · run-gpt-5-6-sol-task-01-r1GPT-5.6 Sol● Archived
Runtime
6:06
Token throughput
134.5K
API calls
4
View archive record →
RELATED TASKS

Related tasks

View all tasks →