I asked ten AI models for the same landing page

On October 20, 2026, I tested AI models with the same one-line brief: "Make the landing page for Ledgerline, an invoicing tool for small design studios: send invoices, automatic polite reminders, see what's owed, bank sync. Return one complete HTML file with inline CSS." I specified the audience, but left the style and sections open. Each model got two tries. Qwen returned nothing usable, and two other tries failed. I ended up with eighteen pages from ten models across nine labs: Claude Haiku 5.5, GPT-6 Luna, GPT-6.1 Sol, Gemini 4 Argon, Grok 4.7, DeepSeek V4.1 Flash, GLM-5.3, MiMo V2.6, Muse Spark 1.3 and Nemotron 3.5.

What the pages looked like

For a while the joke was that every AI-built site had a purple gradient, a dark background and glowing cards. None of these did. One page had a violet button. The rest shared a new look: a warm off-white or white page, a headline with one word in italic or another colour, a small label above it, and a made-up product screen next to it, with rows of invoices and dollar amounts.

I wrote a small checker for these habits. It reads the page's code and styles with plain rules; no model judges anything.

An AI-built page matched a median of 5 of the 8 habits. Eight sites designed by people (Linear, Stripe, Basecamp, Notion, Figma, Are.na, rauno.me and Apple's iPhone page) matched a median of 1.5. Stripe scored highest among those eight, at 4. I don't know whether these models learned from Stripe specifically.

Comparing the sections

Then I cut every page into its parts and asked Claude Code to put them side by side. I wrote "is there a way to see each comopnent across models in matrix or matrixes or in rows if it is easier to see but i would be intesed in componenet level too". It made a matrix with one row per page and one column per part: navigation, hero, feature grid, steps, testimonials, pricing, FAQ, closing call to action, footer. When I saw it I wrote "add the human-designed sites as rows too". The live sites wouldn't split cleanly by code, so those eight were cut by hand.

Here are four of the columns, every page in its own row. Hatched cells are parts the page doesn't have.

And four more columns, the other way round:

The whole matrix, with all thirteen columns, is one large image. There is also an interactive version, where you can pick one part, say the pricing section or the hero, and see every version of it stacked at full width.

The brief never mentioned pricing. Fourteen of the eighteen pages added a pricing section anyway, with plans and monthly prices for a product that doesn't exist.

Eleven pages had testimonials, and every quote was invented, with invented names and studios. Eleven had numbered steps. None of the eight human sites had numbered steps.

The human sites had parts none of these models made: a changelog, a letter from the founder, customers named with their own stories, a page explaining why there are no ads. Those sections were about the actual company and its customers. My brief didn't give the models that kind of information.

Which versions the models preferred

I took a plain page and added one habit at a time, then showed eleven models the plain page and the changed one, 660 pairs in all. For each pair they said which one they would ship, and how likely each was made by AI.

They rated the changed page as more likely to be AI-made for 13 of the 14 habits. But they still often preferred additions like numbered steps or an FAQ. They chose additions like a gradient headline or a glowing button much less often.

Numbered steps were the one habit they didn't rate as more likely to be AI-made, and they chose them every time. The habits that appeared more often in the generated pages also tended to be the ones the models preferred. Across the habits, the rank correlation was 0.63. That makes me question how much I should rely on a model to review these design choices.

Does the date change their choices?

While we were talking about this I wrote "Well web is all designing humans hence humans taste and it changes in time", and then "Sure but can you test micro tastes in recent".

So we collected dated reactions to the dark, glowing look Linear made popular: designers admiring it in early 2023, people tired of the copies by December 2023, and in 2026 calling it the AI-made look. Three models judged the same pair of pages, once told only the date and once given what people said at that time.

Giving them the quotes changed their choices much more than giving them the date alone. With the later reactions, their preference for the dark, glowing page fell to zero. People disagreed more than that. In a separate test, I gave the models a December 2025 reader poll where about 80% said they liked iOS 26. The models still rejected clear Liquid Glass every time. That poll was about iOS 26 overall, so it doesn't tell me how people would choose between these particular designs.

Giving them more than two choices

Asking "which one would you ship" pushes everything to yes or no. I wrote "It sounds like to me we need to anchor it to words we can agree on which is not how we define ourselves. I think we need to define a grey zone not one lense".

We made seven lock screens, from clear glass to frosted, and seven landing pages, from plain to the full dark glow. For each version you say ship it, live with it, or push back. I wanted to see which versions someone would accept, even if they weren't their first choice.

The format mattered more than I expected. Shown one version at a time, the models said "live with it" to almost everything. Shown all seven in order, they pushed back on whatever was at the right end, even when I reversed the order. We then shuffled the set so every version appeared in every position once. That gave us a range without always putting the same design at the end. The choices also changed depending on what was behind the glass.

Apple also changed the opacity during the iOS 26 betas, with more opacity over busy content and more transparency in dark mode. In iOS 26.1 it added a choice between Clear and Tinted.

Trying different personal backgrounds

I also asked the models to choose from two different backgrounds: someone who first met screens around 2000 to 2003, and someone who first met them around 2013 to 2016. I wrote "And maybe we need 3 close points so we can find an average", so we tried three slightly different first-person memories for each background: a blue iMac in a bedroom, a sister's old iPhone 5s. The memories showed up in the words ("the soft candy colors take me right back to fighting over that tangerine iMac") but not in where the models drew the line. In four separate tests, the background changed what the model said, but not what it chose.

Earlier I had written "For example I like Gorillaz. Probably I will see the same crown I saw at a Gorillaz show at camp flag gnaw. Even though I am 41 and I think am I cool and hip, eventually this is our time and space and nothing cool about it from inside". I was thinking about how my taste comes from the people, music and things I've spent time with. Giving the models a description of that kind of background changed their wording in these tests, but not their choices.

What I don't know yet

I tested one brief, for a small software product. Other briefs may bring out other habits. I haven't compared the models' bands with real people's yet; I made a page for that, called Draw Your Band, and haven't run it with people. And I don't know how long this season's look will last. The checker is dated October 2026 for that reason.