Vibeleaderboard
← All Intel
Intel / article

I gave Opus 5.5 one prompt and six hours to visualize Invisible Cities

Source
quesma.com
Author
stared
Date

Context

Piotr Migdał of Quesma gave two coding agents the same one-shot prompt: build a three.js visualization of all of Italo Calvino's Invisible Cities, using up to six hours to make it a masterpiece. In his write-up, Anthropic's Claude Opus 5.5 in Claude Code ran 1 hour 25 minutes with six parallel subagents for about $74 in API cost, and he says it left him mesmerized.

OpenAI's GPT-6 Astra in Codex took 53 minutes for about $10. It worked end to end, but he saw some "AI design slop," with unneeded concepts and comments adding visual noise. The post describes one run per model and the judgment is his, so this is an anecdote about one task, not a . His own takeaway is that AI will be a widely used design tool that burns a lot of .

Terms in this piece · Glossary
  • benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
  • token budget — A cap on how many tokens a task, session, or agent run may consume — the practical control on both cost and how long an agent will grind.
More from stared
Recommended reads
Comments

Checking sign-in…

Loading comments…