Built by Pat
I test AI models by building things. Open each model’s version, copy the exact prompt, or watch the test.
15 builds shown
A rebuild of the morphowater.com product site with its 3D can and scroll motion, using no assets taken from the original, following my clone-app-pat-pro playbook.
A rebuild of the oryzo.ai site with its 3D and motion, using no assets taken from the original.
A rebuild of heronaiapp.com, the same Awwwards site for every model, following my free Clone App Pat Pro skill, with every image made by the model and nothing taken from the original.
Scenes from Moby Dick as a single web page printed like a risograph poster, where every sailor, wave and ink dot is drawn by the model's own code with no images at all, in the same format as Kevin Ngo's a small light.
A test of how closely three models could copy Mana Yerba Mate's moving product page.
Each model chose an Awwwards winner to copy, so this test compared their website choices as well as the finished pages.
A pixel-for-pixel rebuild of an Awwwards winning site that the model picks for itself.
A product page for a mechanical keyboard whose 3D centerpiece is built in code and driven by scroll.
Each model picked its own award-winning site, read the real computed styles off the live page, and rebuilt from those values until every assertion passed.
A website of their choosing, built to show off design and taste.
A website of their choosing, built to show off design and taste.
A scroll-driven photoreal 3D site built around the real IMAX Keighley camera, with an interactive shoot-the-scene finale.
Each model designs an original, award-worthy website, judged on pure taste.
The lightest graphics build (a scroll-driven product site for a film camera that disassembles into an exploded view as you scroll) went to GLM, and cheaply. Here is that prompt:
Then a design showcase site, total creative freedom, meant to show off peak taste. Both models, given a blank canvas, independently reached for the deep ocean. GLM made NOCTILUCA, a bioluminescence field study; Kimi made ABYSSAL, a vertical survey of the deep. Here is the prompt:
Pat Simmons gave AI models the same builds, three to five at a time, and logged every run's cost, tokens and minutes. DeepSeek V4.1 Flash cost as little as $1.00 a build and Fable 5.1 as much as $592.37, and the two cheapest models placed low in the blind rankings. The page also lists every model's API price and the main plans, checked November 10, 2026.
Fable 5.2 costs twice as much per token as Opus 5.5. We gave both the same email thread, 48-page report, messy contact list, slide deck and small build. Three jobs were a tie, Fable won two on details, and it cost about twice as much to get there.
ChatGPT is OpenAI's all-in-one app, free or from $8 a month, with GPT-6 Astra on Plus and Pro. Qwen is Alibaba's free assistant and a family of open models you can download. Plans, sizes, licenses and Alibaba's own scores, from each company's pages.
I put Google's new frontier model through four hard builds against Opus 5.5, Fable 5.1 and GPT-6 Astra.
Turns a plain-English video idea into a ready-to-paste prompt for the Seedance 2.0 video model.
Lets Claude Code build and play 3D cities in Unreal Engine 5.8 through the official Unreal MCP plugin.
AI FOR MORTALS.
Every weekday, the practical AI updates that matter, and exactly what to do with them.
The latest in AI, in plain English. Free, every weekday.