A finished website in a median of 9.6 seconds
One sentence describing a business goes in. Three complete, personalized websites come back, live at public addresses. Measured on production, 16 to 17 August 2026: fastest run 8.53 seconds, median 9.6 seconds, slowest run 17.18 seconds, across eight runs. No runs discarded.
Every run we made
| Prompt | Generate | Render | Total |
|---|---|---|---|
| Dental clinic accepting new patients | 7.47s | 1.06s | 8.53s |
| Italian restaurant in downtown Toronto | 8.14s | 1.13s | 9.26s |
| SaaS analytics platform for ecommerce founders | 8.40s | 0.99s | 9.39s |
| Handmade ceramics online shop | 8.23s | 1.32s | 9.54s |
| Dog grooming salon in Vancouver | 8.42s | 1.31s | 9.73s |
| Wedding photographer in Seattle | 8.88s | 1.33s | 10.21s |
| Law firm specialising in immigration | 11.01s | 1.24s | 12.25s |
| Yoga studio in Austin, Texas | 15.54s | 1.64s | 17.18s |
How it was timed
Timing starts the moment the prompt is submitted and stops the moment the finished website is fully rendered and reachable at a public address. Generate covers writing the page copy and matching designs; Render covers producing the finished HTML. Both are included in the total. Run against live production from a standard connection, with no warm-up, no cache priming and no retries. Each run returned three complete designs, of which one was timed to render.
What this does not mean
An AI is not writing a website from scratch in ten seconds. One AI call writes the copy for your business. The layouts are designs made by human designers, selected by matching the prompt against a catalogue and then personalized with that copy. That division is exactly why it is fast, and why the result does not look machine-generated.
It is not always under ten seconds. Five of the eight runs were. One took 17.18 seconds. The generation step dominates the total and depends on an external model provider, so it varies.
Not every design is a full page. Some designs in the catalogue are single-screen heroes rather than multi-section sites. One of the eight runs returned one.
These were measured by us, which is why the method is written out in enough detail to repeat. Run it yourself and tell us what you get.
For comparison
Our own from-scratch generative pipeline, where an AI writes the site code rather than personalizing a prepared design, averages between three and six minutes across 252 real customer builds. That is consistent with the rest of the category. The method on this page is roughly thirty times faster than that.
We have not independently timed every competing product, and we would rather publish nothing than publish a competitor's number we cannot stand behind. If you time one against the same method, send it to us and we will add it, whichever way it goes.
Product
Who It's For
- For Agencies
- For Freelancers
- For Hosting Providers
- For SaaS Platforms
- For Startups
- For Financial Services
- For Healthcare
Features
Resources
- Blog
- Help Center
- Free Business Tools
- Free UI Components
- Website Templates
- Template Collections
- Website Examples
- Guides
- Comparisons
- Alternatives
- Best Lists
- Reviews
- Pricing Guides
- Use Cases
- Glossary & Guides
- Integrations
- Developers
- Benchmark
- Windows App