Claude Fable
Anthropic's most capable model at its time of release, positioned as representing an entirely new model category and competing directly with the GPT-5.6 series.
Core Facts
Timeline (last 90 days)
Tim believes that in 99% of cases Opus should be preferred over Fable, because Fable's additional gains are not enough to offset the cost.
Willison had previously used Claude Fable with P5.js to simulate watercolor illustrations.
On KernelBench (low-level GPU code writing), DeepSeek C.E Flash 1017 ranks third, behind only the latest Claude model, but beats Gemini 4 Argon, Claude Fable and GPT Leo Astra.
The Claude Fable version shows a slight improvement in response conciseness compared to the Opus version, but the difference is not significant.
All Facts (5)
The Claude 3 series uses Haiku, Sonnet, and Opus to name different capability tiers
65%UnverifiedTim believes that in 99% of cases Opus should be preferred over Fable, because Fable's additional gains are not enough to offset the cost.
50%UnverifiedWillison had previously used Claude Fable with P5.js to simulate watercolor illustrations.
50%UnverifiedOn KernelBench (low-level GPU code writing), DeepSeek C.E Flash 1017 ranks third, behind only the latest Claude model, but beats Gemini 4 Argon, Claude Fable and GPT Leo Astra.
50%UnverifiedThe Claude Fable version shows a slight improvement in response conciseness compared to the Opus version, but the difference is not significant.
50%