{
  "id": 4972565,
  "title": "Claude Fable 5.1 made me a really nice animated pelican",
  "url": "https://urgent.news/2026/09/01/claude-fable-5-1-made-me-a-really-nice-animated-pelican",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-01T23:57:28.000Z",
  "source": {
    "name": "Simon Willison",
    "slug": "simon-willison",
    "url": "https://simonwillison.net/2026/Sep/1/claude-fable-5-1/"
  },
  "original_language": "en",
  "account": "On September 1st, 2026, Simon Willison published an article titled \"Claude Fable 5.1 made me a really nice animated pelican\" on his website. In the piece, Willison shared his experience with Anthropic's Fable 5.1 model, which they claim sets a new standard for coding, knowledge work, and long-running problem-solving tasks.\n\nWillison began by noting that the model achieved a 52.6% score on the brand new Terminal-Bench-Science 0.1 benchmark, which was a significant improvement over previous models. He then proceeded to run a series of prompts through Fable 5.1 at various reasoning levels to observe the output.\n\nFor the low reasoning level, Willison found that Fable 5.1 appeared to skip reasoning entirely and produced an SVG of a pelican riding a bicycle. The output was generated in 23.8 seconds for a cost of 10.017 cents. Interestingly, when he ran the same prompt at the medium reasoning level, the output token count was 21 tokens less, indicating that Fable 5.1 may have skipped reasoning at this level as well.\n\nAs the reasoning level increased, so did the output from Fable 5.1. At the xhigh level, the model took 7 minutes and 51 seconds to generate an SVG with 65,927 output tokens. This version included a pelican with detailed features such as a long neck, orange beak, and intricate wings, as well as a bicycle with realistic elements like spokes, wheels, and handlebars.\n\nAt the max reasoning level, Fable 5.1 produced the most impressive result, taking 13 minutes and 54 seconds to generate a 65,927 token SVG. This version featured added details such as a darker tip region representing the primary feathers, a more natural-looking trail edge, and a helmet for the pelican to match the bicycle theme. Willison noted that while the output was still not as detailed as some other models, it was still an improvement over previous iterations.\n\nThe article concludes with Willison pondering potential additional enhancements, such as adding a scarf or cap for the pelican, or adjusting the handlebar curve for a more realistic appearance. He also mentions a discussion on Hacker News about the possibility of creating an animated version of the pelican, which he chose not to do due to additional costs.",
  "summary": "Today is Claude Fable (and Mythos) 5.1 day . Anthropic say that Fable 5.1 \"sets a new standard for coding, knowledge work, and long-running problem-solving tasks\". Their announcement spends a notable amount of time on scientific research, boasting of a 52.6% score on the brand new Terminal-Bench-Science 0.1 benchmark (first announced on August 27th ), up from 24.7% for Fable 5, 29.0% for Opus 5…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}