Gemini 3.5 Pro drops today. Perfect timing.
๐ฅ Just recorded a full tutorial showing how to build a $10,000 product website with Google AI Studio + Higgsfield (step-by-step, 0 coding)
Even the product and its film are AI-generated.
Made this with Opus 5 in one sitting. Quick model. Honest truth though, the same prompt lands on Fable or Kimi too. The spec does the work, not the logo on the box.
@rileybrown The 37-minute research run is the real unlock. It's not the output, it's that it actually reads everything first instead of guessing. Opus 5 leaning into patience is underrated.
@jameslaurents The offset depth on that revenue card carries the whole layout. Bento only reads this clean when one tile breaks the grid instead of every tile being equal. Nice set.
@Av1dlive Once you can run any of them, the prompt is the real variable. Fed the same design spec to a few models and got basically the same site. Recorded the whole process here
https://t.co/DkQXLW2fu2
You can build an Awwwards-level site on almost any AI model right now. The model isn't the hard part. The prompt is. So I recorded the whole process. Copy one prompt, paste it into Claude, and you get a cinematic 3D scroll site in about 10 minutes.
Full tutorial ๐
@Oluwaphilemon1 The wild part is the model barely changes the final site when the prompt is tight. I ran the same spec across a few and got near-identical results. Did a full walkthrough here if it helps
https://t.co/DkQXLW2fu2
You can build an Awwwards-level site on almost any AI model right now. The model isn't the hard part. The prompt is. So I recorded the whole process. Copy one prompt, paste it into Claude, and you get a cinematic 3D scroll site in about 10 minutes.
Full tutorial ๐
@markproduct 5sdesign โ a library of ready-to-copy prompts, each with its own design system, for building sites with any AI model. Just posted a full build tutorial here:
https://t.co/DkQXLW2fu2
You can build an Awwwards-level site on almost any AI model right now. The model isn't the hard part. The prompt is. So I recorded the whole process. Copy one prompt, paste it into Claude, and you get a cinematic 3D scroll site in about 10 minutes.
Full tutorial ๐
It's all at 5sdesign. Ready-to-copy prompts, plus design systems, backgrounds, fonts, colors, and icon packs for whatever model you build with. https://t.co/evP8c5Cbzo
You can build an Awwwards-level site on almost any AI model right now. The model isn't the hard part. The prompt is. So I recorded the whole process. Copy one prompt, paste it into Claude, and you get a cinematic 3D scroll site in about 10 minutes.
Full tutorial ๐
Opus 5 drops Thursday and half of you are about to rewrite every prompt again.
Kimi K3 last week. Sol before that. Fable before that.
The model on top changes every week. Your design spec doesn't have to. ๐ฎโ๐จ
@aidenybai Most of that "not production ready" is just no constraints imo. Hand it a real design spec and Kimi's frontend gets way closer. Benchmark makes the gap look bigger than it is
@pankajkumar_dev Opus 5 Thursday. Kimi K3 last week. Sol before that. The model on top keeps changing โ a design spec that runs on all of them doesn't. Models are rentals. The spec is the asset.
Careful is the production bar, agreed. But careful can live in the spec too โ hand any model a real design system and it stops improvising. The weights matter less when the constraints are tight.
Fable is careful. None of the other models are careful. GPT-5.6 Sol, Opus, Kimi, Grok. You can compare them all day long on capabilities, and it doesn't matter, because in order to use a model for real production work, it must first and foremost be careful.
That's the only dimension that matters to me. And if you're doing real production-facing customer-facing work with AI, it's the only thing that should matter to you, too. The other models will not be truly competitive at anything but proofs until they are trained to be careful.