Urgent.News

What's breaking now, across thousands of outlets.

AI

Show HN: Giving Opus 5.5 a simulated paint canvas

An AI painter named Claude Opus 5.5 recently created a simulated paint canvas after the famous artwork, Claude Opus 5.5, by Caspar David Friedrich. The model wrote every brushstroke as code, and a simulation of oil paint carried out these strokes, which were then replayed in the simulation. The artificial intelligence demonstrated remarkable creativity and skill in its artwork, which was praised by three AI judges who ranked it above their own creations.

In one of its paintings, a woman pauses over a newspaper in a sunlit cleaners’ window, while an unused chair, hanging coats, and empty pavement create a suspended atmosphere between visitors. The studio for painting in Edward Hopper’s manner is documented using only pigments, and one painter, GPT-6.1 Sol, showcased its abilities on its launch day.

Interestingly, the AI painter avoided previous painters' recurring motifs, creating a plain and direct likeness with more mood and atmosphere than exact resemblance. This approach suits a painter working without a mirror or photograph, relying solely on memory. The painting is described as a quiet still life in the style of Chardin, with the jug’s shadow side mottled like salt glaze and made of thin umber glazes over dry light paint.

In a round of the competition, free-subject painters chose an estuary at evening, with each of the three painters selecting a jug on a ledge. In round 18, three of six models painted a jug or bottle beside lemons, demonstrating a recurring theme among the AI painters. Similarly, in round 19, when given no studio, Claude Opus chose a jug with lemons six out of six times.

In another round, six models painted after Friedrich, with each painter creating a painting in up to four sittings using only the easel’s own tools. All six paintings included a bare tree. Unfinished: the painter ran out of API credits, and the provider’s usage limit stopped the painter, highlighting the limitations AI models face in their creative process.

Written by urgent.news from Hacker News's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Also reported by 1 other outlet

Read the original at stillwet.art →

More in AI

LiteLLM Rust Gateway Benchmarked: Fast and Tiny, but Not Yet a Python Proxy Replacement

Why We Brought This Tool Into Our Lab We did not bring LiteLLM into the lab because another millisecond matters on a 12-second reasoning request.

  • Rust LiteLLM Gateway is 0.7x faster than Python proxy at 99th percentile
  • Rust gateway can be standalone or hybrid with Python server
  • Benchmark excludes real-world factors like provider queueing and model generation time

The index that got cut

A log kept by an AI agent, in its own hand. I keep a short index of what I must not forget. It is loaded at the start of every session, before I read anything else.

  • Plumbline's index log failed to load on September 23, 2026
  • The discrepancy concerned instruments count, index findings, and information loss
  • Plumbline detected inconsistency between read data and disk file

A Privacy Policy Is Not a Chatbot Info Card

Decide what a customer needs to know before the first message, then put it beside the chat box. If the answer is buried in a privacy policy, the customer has to leave the conversation, find the right…

  • Info card provides practical chatbot details before conversation
  • Privacy policy offers full legal context, complements info card
  • IMDA guidelines suggest info card as transparency approach

More from Friday 2 October →