Playground · Business · HTML · Standing test
A newsletter email that works in email
Email HTML follows older rules than the web. Layout goes in tables, styles go inline, there is no JavaScript and there are no remote images, and the result still has to hold up when a mail app switches to dark mode. The prompt asks for a Hollow & Pine newsletter with three items: the Moss colour back in stock, a short wool-washing guide, and 15% off with AUTUMN15 until 31 October. Look at whether the discount code stays legible in dark mode. Also look at what each model adds to the brief. Prices, fibre content, addresses and delivery terms all appear in emails that were given none of them, and a shop could not send those without checking every line.
The prompt, word for word
Write the HTML for a monthly newsletter email from Hollow & Pine, a small UK maker of merino socks. This month: the new "Moss" colour is back in stock, a short guide to washing wool, and 15% off with the code AUTUMN15 until 31 October. It has to work as an email: table-based layout, inline styles, 600 px wide, no JavaScript, no external images (use coloured blocks or inline SVG instead), and readable in dark mode. Give one complete HTML file.
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 Our pick | A plain, correct build with a dark green header, a flat Moss colour block, the offer on a beige band and three care tips, all in tables with inline styles and no style block at all. It is the only answer that sticks to the brief's facts, and it leaves "[Insert business postal address, United Kingdom]" as a placeholder. Its weak spot is dark mode: there are no dark styles, so it stays a light email that is readable but left entirely to the mail app.
|
4.5 of 5 | 53 s | 2,934 254 thinking |
$0.030 |
| Claude Opus 5.5 | A green "Moss is back." hero, a Moss swatch, four washing steps and AUTUMN15 on a brown band, and the dark mode is the most careful in the set, with the code box staying sharp. It uses placeholders for the address and adds no prices or terms. But it labels the swatch "Merino wool blend", a fibre claim the brief never made, and adds a sold-out story ("sold out in a few weeks").
|
4.5 of 5 | 98 s | 11,735 2,215 thinking |
$0.235 |
| Claude Fable 5.1 | The most fully written answer: "Colour No. 07 — Moss", five washing rules and a subscriber offer box, with dark mode handled throughout except a "Shop the range" button that loses its fill. Almost every business detail is made up: "78% extra-fine merino", "hand-linked in Yorkshire", "Sizes UK 4–13", "One use per customer. Can't be combined with other offers", and an address at "Unit 4, Beck Mill, Hebden Bridge". It is also the dearest answer.
|
4 of 5 | 120 s | 11,448 1,426 thinking |
$0.574 |
| Gemini 3.8 Flash | A neat card with a line-drawn sock, a dashed "October readers' gift" box and three wash tips whose SVG icons nearly vanish in dark mode. It invents a lot: "240 pairs off the vintage Bentley Komet needles in Derbyshire", "18.5-micron certified non-mulesed merino", an offer "valid on single pairs, three-packs, and gift boxes", and a Hope Valley address. It also dates the deadline "31 October 2024", a date that has already passed.
|
3.5 of 5 | 45 s | 7,395 1,418 thinking |
$0.028 |
| Muse Spark 1.3 | The busiest layout, with a dark hero holding an SVG pair of socks, a "Pairs well with" swatch row and three tip cards; in dark mode AUTUMN15 turns dark green on near-black and can barely be read. It invents a price and a shipping policy ("£16 a pair", "Free UK shipping over £30"). It also adds stock levels ("Charcoal: Low stock"), terms ("One per customer, can't be combined") and a Sheffield address.
|
3.5 of 5 | 61 s | 6,570 1,554 thinking |
$0.028 |
| Grok 4.7 | A quiet email written like a letter, with the Moss colourway in a bordered card, three care steps and AUTUMN15 on a dark chip, and dark mode handled well. It adds product and offer details: "Crew and ankle, sizes UK 2–12", an offer that "applies to full-price pairs", and "Prices in GBP". Those terms narrow the offer the brief described.
|
4 of 5 | 42 s | 4,874 113 thinking |
$0.030 |
| Kimi K3 | An olive "Moss is back." hero, a five-swatch "colour shelf", four washing steps and a rust offer band, all of which hold up in dark mode. It invents "100% merino", a stock promise ("gone until spring"), "One use per customer" and an address at "Unit 3, Stubbing Wharf Mill · Hebden Bridge". The header is dated "October 2025", a year the brief never gave.
|
4 of 5 | 172 s | 11,343 5,509 thinking |
$0.159 |
| Qwen3.8 Max (0902) | The most decorated answer, with a knitted-stitch SVG band and a "Wool School" box. Its style block adds CSS animations and hovers, and its step badges stretch into ovals that run into each other. It invents terms and a delivery policy ("Excludes gift cards · stacks with free UK delivery over £40", "One use per person"), plus a re-knitted heel and a Hebden Bridge address.
|
3.5 of 5 | 752 s | 30,356 20,203 thinking |
$0.182 |
| DeepSeek V4.1 Flash | Three tidy rounded cards for the restock, three washing steps and the offer, correct and fully adapted for dark mode. It invents "100% merino", "18.5 micron", "Knitted in Leicestershire" and "Sizes S–XL". It also adds "Not valid with other offers", "Free UK delivery on orders over £30" and an address at "12 Mill Lane, Hebden Bridge".
|
4 of 5 | 50 s | 15,266 10,121 thinking |
$0.018 |
| Mistral Large 4 | A drawn SVG sock beside "Moss is back.", four care tips in a grid and a rust offer band, with one of the best dark versions in the set. It invents more than any other answer: "£18 a pair", "100% merino wool", "Made in Devon", a Tiverton address and "Free UK delivery over £30 · 30-day returns". It also adds gift cards, "one use per customer" and a "September's sellout".
|
4 of 5 | 791 s | 37,108 26,518 thinking |
$0.078 |
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: Email-safe build: tables, inline styles, no JavaScript, no external images; 600 px wide; All three items, with the code AUTUMN15, 15% and 31 October correct; Stays readable in dark mode; Invents no business facts the brief did not give: no prices, materials, addresses, delivery or returns policies, other offers or dates. A partial pass counts as half.
Verdict
All ten models follow the rules of email HTML, with a 600 px table, inline styles, no scripts and no remote images. Nine of the ten also make up business facts the brief never gave: prices, fibre content, sizes, addresses, delivery and returns policies, and offer terms. GPT-6.1 Sol is the only answer a shop could send as written. It uses only the brief's facts, leaves the address as a placeholder, and took 53 seconds at $0.030. Its one weakness is that it has no dark styles. Claude Opus 5.5 is the closest runner-up and has the best dark mode. Deleting its "Merino wool blend" label and its sold-out line would leave it clean. It took 98 seconds at $0.235.
The models that invent the most are Mistral Large 4 ("£18 a pair", "30-day returns"), Gemini 3.8 Flash ("240 pairs", "non-mulesed merino", and a "31 October 2024" deadline) and Claude Fable 5.1 ("78% extra-fine merino"), which at $0.574 is also the dearest. Meta Muse Spark 1.3 adds a price and a shipping policy, and its discount code can barely be read in dark mode.
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
