Claude Fable 5 outpaces GPT-5.5 by 13 points on FrontierMath's toughest problems
Anthropic's Claude Fable 5 achieves 88% accuracy on FrontierMath's hardest tier, a dramatic jump from Claude Opus 4.5's sub-10% performance in early 2026
Published 8sem1 sourceNotable
Lire en français
≈ 26s
10 %
5's sub-10% performance in early 2026
75 %
5 reaches approximately 75% on the same benchmark, 13 percentage points behind Claude Fable…
The fact
OpenAI's GPT-5.5 reaches approximately 75% on the same benchmark, 13 percentage points behind Claude Fable 5
AI models' mathematical reasoning capabilities are accelerating notably, intensifying competition in advanced problem-solving benchmarks
Click the link to read an article on the topic: