Both @OpenAI and @Claude by Anthropic launched their frontier coding models last month, and @sama posted:

we still get looksmaxxed on frontend a little but we IQmog hard now

In other words: GPT-5.5 has the brains; Claude still has the taste.

To verify this statement, the @Kilo Code team gave both models the same 5 UI design prompts - a SaaS landing page, an analytics dashboard, a settings page, a sign-up screen, and a pricing page.

TL,DR: Yes. Opus 4.7 produced more polished UIs. Its outputs feel closer to a real product. GPT-5.5's outputs work and look okay, they still feel more like Tailwind UI starter templates than custom designs.

Key takeaways

• Neither output would be ready to ship at first shot. Both would require more work to fix the details, refine spacing, and polish components.

• Claude Opus 4.7 shows better understanding of typography, color, spacing, and UI details.

• The per-token pricing suggests GPT-5.5 is the most expensive model ($30 per 1M output tokens vs. $25); In practice, Opus 4.7's longer reasoning used more tokens tho.

Read the full breakdown here →

GPT-5.5 has clearly improved at UI work. For rapid prototypes where visual choices don't matter much, it feels like a solid option.

Would you switch?