llm-anthropic adds Fable 5.2 and Mythos 5.2 support; max effort fails for pelican test
Overview
Simon Willison has added Fable 5.2 and Mythos 5.2 support to llm-anthropic, so users can run prompts through them with the llm tool using the model ID claude-fable-5.2.
Willison ran his pelican-on-a-bicycle SVG prompt at five reasoning levels: low, medium, high, xhigh and max. At max, the model returned no image after using its full 128,000 thinking tokens. He says the same max-effort failure occurred with Opus 5.5, while Fable 5 and Fable 5.1 returned max-effort pelicans correctly.
Written by AI from the articles below · updated Nov 9, 7:41 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Simon WillisonBlogSimon Willison adds Fable 5.2 and Mythos 5.2 support to llm-anthropic 0.31
AISimon Willison has added Fable 5.2 and Mythos 5.2 support to llm-anthropic, letting users run prompts through them with the llm tool and the model ID claude-fable-5.2. He ran pelican SVG prompts at five reasoning levels, low, medium, high, xhigh, and max, and max returned no image after exhausting 128,000 thinking tokens. He says the same max-effort bug appeared with Opus 5.5, while Fable 5 and Fable 5.1 returned max-effort pelicans correctly.
Heat Trend
Not enough continuous observations to show a trend yet.