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

AI models can now generate complex interactive visualizations from a single prompt, potentially transforming design workflows for creators and developers.

The author, Piotr Migdał, tested Claude Opus 5.5 by tasking it with creating a three.js visualization of Italo Calvino's "Invisible Cities." Migdał provided a specific prompt and a six-hour time limit, instructing the AI to treat it as a one-shot task and aim for a masterpiece. While Opus 5.5 indicated it used roughly half the allotted time (1 hour 25 minutes) with six subagents running in parallel, the author found the result impressive, despite the API cost of $74. For comparison, GPT-6 Astra completed a similar task with the same prompt in 53 minutes at a cost of $10, but Migdał subjectively found Opus 5.5's output to be of higher quality, aligning with his belief that newer models represent significant jumps in design capabilities.

Claude Opus 5.5's performance on generating an interactive visualization of "Invisible Cities" suggests a significant leap in AI's design and complex task execution capabilities. The model's ability to produce a high-quality output within a fraction of the allotted time indicates growing efficiency and sophistication in AI-driven creative tools. This advancement positions AI as a more potent partner in design and interactive media creation, challenging traditional workflows and raising questions about the evolving roles of human creators.

The comparison with GPT-6 Astra highlights a competitive landscape where AI models are rapidly improving in both technical execution and aesthetic output. The substantial difference in API costs ($74 for Opus 5.5 versus $10 for Astra) also underscores the economic considerations for users, suggesting a trade-off between cost and the perceived quality or complexity of AI-generated designs. As these models become more capable, expect increased adoption in fields requiring visual design and rapid prototyping, potentially lowering barriers to entry for complex interactive projects.