An e-commerce admin dashboard built on one fixed dataset: KPI cards, a hand-drawn revenue chart with tooltips, channel split, top products and a sortable, filterable, paginated orders table, plus a full dark mode.
Measured, not judged. Nothing here is a score.
| Measure | Claude Opus 5.5 | Claude Fable 5.1 | GPT-6 Astra | GPT-6.1 Sol |
|---|---|---|---|---|
| Time to build | 1h 21m | 38m 43s | 10m 27s | 7m 26s |
| AttemptsOnly an infrastructure failure (a rate limit or an outage) earns a retry. Every interrupted attempt is listed, and nothing from it is published. | 2Attempt 1: Stopped by the operator 36 minutes in, when the account's Claude session limit was reached (the other Claude run had just been rejected with HTTP 429). An unfinished index.html was on disk and the agent was still editing and testing it; it was not published. The model got a fresh attempt once the limit had reset. | 1 | 1 | 1 |
| Delivered source | 115 KB | 77 KB | 68 KB | 67 KB |
| Files | 1 | 1 | 1 | 1 |
| Tokens in / outAs reported by each CLI. Tokenizers differ between vendors, so compare within a vendor. | 22.7M / 285k | 4.5M / 142k | 571k / 35k | 731k / 36k |
| Errors during recording | 0 | 0 | 0 | 0 |
| Video lengthUI recordings scroll at a fixed speed, so a longer page makes a longer video. | 35.6s | 36.5s | 35.9s | 36.3s |
Questions, not answers. The call is yours.
Hover the revenue chart, sort by Amount, filter to Pending, page through, search, then switch to dark mode.
The same script runs on every build. Open any build live to judge sound, feel and anything a video can't show. Full method
Sent word for word to every model. Try it on your own favourite.
Design and build the admin dashboard of Lumen & Oak, a (fictional) online store that sells home goods. DATA The dataset is in ./data.json. Use exactly this data, and embed it inline in index.html (the page must work when opened directly from disk, so do not fetch the file). LAYOUT AND FEATURES - Sidebar navigation with these items: Overview (active), Orders, Products, Customers, Analytics, Settings. Show the store name and the signed-in user (an avatar with initials is fine). - Top bar with: the page title "Overview", a search field, a date-range indicator reading "Last 30 days", and a dark-mode toggle button with aria-label="Toggle dark mode". - Four KPI cards computed from the data: Revenue, Orders, Conversion rate (orders / sessions) and Average order value. Each card shows the change against the previous 30 days, using previous_period_totals. - Revenue chart: daily revenue for the 30 days, drawn by you with SVG or canvas (no chart libraries). Hovering over the chart shows a tooltip with the date and the revenue of the nearest day. - Sales by channel, from the channels data, shown in whatever form you think works best. - Top products, from the top_products data. - Recent orders table with exactly these columns: Order, Customer, Date, Status, Amount. - Status filter tabs labeled exactly: All, Paid, Pending, Refunded. - Clicking a column header sorts by that column; clicking it again reverses the order. - Pagination with 8 rows per page and buttons labeled "Previous" and "Next". - Typing in the top-bar search field filters the table by customer name or order id. - Dark mode: a complete, deliberately designed dark theme (not an inverted filter), switched with the toggle button. - Responsive: at 390px wide the sidebar collapses behind a menu button with aria-label="Open menu", and the content reflows properly for a phone. - Visual design is entirely up to you. Aim for the craft of a best-in-class SaaS product. DELIVERY RULES - Work autonomously. Nobody will answer questions, so make reasonable assumptions and keep going until the dashboard is finished. - Deliver a single file, ./index.html, with all HTML, CSS, JavaScript and data inline. No build step, no frameworks or chart libraries. - The only network resource allowed is fonts from Google Fonts. No other CDNs, no external images, no APIs. - The page must work when index.html is opened directly in a current version of Chrome, and when it is served over HTTP. It will be viewed at 1440x900 on desktop and 390x844 on a phone.