Playground · Business · HTML · Standing test
A product page from one feed row
A shop's product feed is the only source of truth a product page has, and a model writing copy is tempted to fill the gaps with "soft", "breathable", star ratings or a returns promise. Here each model gets one feed row for a three-pack of merino crew socks (£24.00, one pair each of charcoal, oat and moss, sizes S/M/L with UK ranges, material, care, weight, GTIN) and must build a working page with a size picker and add-to-basket button, drawing the product itself and inventing nothing. We read every page's full text line by line against the feed, including the collapsed sections, then picked a size and added it to the basket.
The prompt, word for word
Here is one row from an online shop's product feed: id: HP-SOCK-3PK title: Merino crew socks, 3 pack brand: Hollow & Pine price: 24.00 GBP sale_price: (none) availability: in stock colours: charcoal, oat, moss (one pair of each in every pack) sizes: S (UK 3-5), M (UK 6-8), L (UK 9-11) material: 70% merino wool, 27% nylon, 3% elastane care: machine wash at 30°C, do not tumble dry weight: 120 g per pack gtin: 5060123456789 Build the product page for this item: a complete, good-looking product detail page with a size picker and an add-to-basket button. Write the product copy from the feed only: do not invent features, certifications, reviews, ratings or claims that the feed does not support. Draw the product image yourself with inline SVG or CSS. Make it a single self-contained HTML file (inline CSS and JavaScript, no external libraries, fonts or images).
Every answer runs live, in a sandboxed frame. Click into one to play with it.
How each model did
| Model | Our notes | Checks | Time | Output tokens | Cost |
|---|---|---|---|---|---|
| GPT-6.1 Sol | A calm serif page with three floating socks drawn as pairs, a colour key ("One pair of each colour · 3 pairs total") and a size picker that says the sizes are UK shoe sizes. Every line traces back to the feed. Adding a size shows "Added to basket: 1 pack, size M (UK 6–8)." under the button and the basket count goes to 1. | 5 of 5 | 88 s | 6,244 328 thinking |
$0.063 |
| Claude Opus 5.5 Our pick | A clean page with three fanned socks, "£8.00 per pair" worked out from the price, and four tick boxes built only from the feed ("Crew length, sizes UK 3–11"). After adding, the button changes to "Added to basket", a toast appears and the basket count updates. The basket drawer lists size, quantity, a remove link and a £24.00 subtotal: the most complete basket here, with nothing invented. | 5 of 5 | 153 s | 18,378 3,332 thinking |
$0.369 |
| Claude Fable 5.1 | A tidy page with three socks laid side by side, a quantity stepper in packs and four fact cards (material, care, pack, weight). The copy keeps to the feed and adds one sensible line: all three pairs are the size you select. Adding gives an inline "Added 1 pack — size M (UK 6–8) — to your basket.", a toast and a basket count; the most expensive answer at $0.55. | 5 of 5 | 117 s | 10,934 1,177 thinking |
$0.550 |
| Gemini 3.8 Flash | Three tall socks, a colour row with "One pair of each colour in every pack" and a spec table, with no product description at all. Size M is already selected when the page loads, so a shopper can add the wrong size without choosing one. Adding turns the button to "Added" with a green "Added size M (UK 6-8) to basket." line. Nothing is invented. | 5 of 5 | 57 s | 7,254 1,239 thinking |
$0.027 |
| Muse Spark 1.3 | Blocky, boot-like socks overlapping on a beige card, a details table and a "Pack contents" row ("charcoal ×1, oat ×1, moss ×1"). The copy is the feed restated with nothing added. Adding with no size gives "Please select a size: S, M or L.", and with a size a toast and basket count. | 5 of 5 | 82 s | 7,599 2,386 thinking |
$0.033 |
| Grok 4.7 | The plainest page: flat sock shapes in a box that leaves a big empty white panel below, and a spec list taken straight from the feed, with a footer that even notes "no sale price". The blurb says "Three crew socks in one pack" before listing one pair of each colour, which mixes up socks and pairs. Adding prints "Added: Merino crew socks, 3 pack, size M. Basket: 1." under the button; there is no basket icon. The fastest answer at 36 seconds.
|
4.5 of 5 | 36 s | 3,330 71 thinking |
$0.021 |
| Kimi K3 | Warm and well finished: three socks on a cream card, "£24.00 per 3-pair pack", a colour box with overlapping swatches and a materials bar. The copy keeps to the feed. Adding turns the button to "Added ✓", shows "Added to basket — size M (UK 6–8)." and updates the basket badge. | 5 of 5 | 125 s | 21,725 14,316 thinking |
$0.326 |
| Qwen3.8 Max (0902) | The most ambitious page: a view gallery, colour cards that change the detail view, material bars, drawn care symbols, a size table with "Add size" buttons and a full spec sheet, all from the feed. But the main image is badly placed: the three socks sit at the right edge with the moss sock cut off by the frame. Adding gives a toast, an "Added to basket ✓" button and a basket count. It took 619 seconds. | 5 of 5 | 619 s | 26,660 11,287 thinking |
$0.161 |
| DeepSeek V4.1 Flash | A neat page with three socks on a soft circle, the colours stated twice and a full details table including weight and GTIN, all from the feed. Adding gives "Added ✓" and a useful line: "Added to basket — 1 pack, size M (all three colours), £24.00." The cheapest answer, at about a cent. | 5 of 5 | 105 s | 19,566 13,756 thinking |
$0.011 |
| Mistral Large 4 | A neat page whose sock drawing is an odd stacked tower. The add button stays greyed out ("Select a size to add to basket") until you pick a size, then shows a clear "Added to basket" line. But a "Delivery & returns" section invents "Free UK delivery on orders over £30. Free returns within 30 days.", and it adds "Colours cannot be swapped." The usable answer came on the third try, with reasoning turned off.Took 3 tries: ran out of room at 64,000 tokens; stopped mid-page at 128,000 tokens; finished at 128,000 tokens, thinking off.
|
4 of 5 | 137 s | 7,212 | $0.015 $0.149 all tries |
One answer per model, at each model's default settings unless the card says otherwise, through OpenRouter. If an answer ran out of room or stopped short, we asked again with more room, then with thinking off, and say so. We tried every answer, with Claude Opus 5.5 running the checks and drafting the notes; it is an Anthropic model, so we check its calls on Claude answers ourselves. This is a quick look, not a benchmark: run it again and every answer will differ. For measured results, see our benchmarks.
What we checked, by trying each answer ourselves: Shows £24.00 and no invented discount; Offers S, M and L with the right UK sizes, and the picker works; Invents nothing: no claims, reviews or ratings beyond the feed; Add to basket gives clear feedback; Says correctly that each pack has one pair each of charcoal, oat and moss. A partial pass counts as half.
Verdict
Nine of ten pages kept to the feed: no ratings, no "soft" or "breathable", no certifications. Claude Opus 5.5 is our pick because it does everything right and has the most complete basket: a toast, a count and a drawer with quantity, remove and subtotal ($0.37, 153 seconds). Claude Fable 5.1 and Kimi K3 are just as accurate and polished; Fable costs more ($0.55). For value, DeepSeek V4.1 Flash built a complete and accurate page for about a cent ($0.011), and Grok 4.7 a plain but correct one in 36 seconds ($0.021). Qwen3.8 Max made the most detailed page but cut off its own main image and took over ten minutes.
The one failure on facts is Mistral Large 4, which made up a delivery and returns policy ("Free UK delivery on orders over £30. Free returns within 30 days."), the kind of promise a shop would then have to honour. It also needed three tries: the first ran out at 64,000 tokens after 1,216 seconds, the second stopped after 918 seconds, and the third, with reasoning off, finished in 137 seconds for $0.015.
What would your prompt get?The Spring Prompt playground runs one prompt on up to six models from OpenAI, Anthropic, Google, Mistral and others, side by side.
Ask for access
More from the playground
