7 days ago, Anthropic announced its new model, Claude Fable 5.2.
And as always,the internet is full of people telling you how amazing it is.
You’ll see:
Posts explaining why it’s a game changer
Screenshots of impressive benchmark results.
Tutorials showing you how to get the most out of it.
And hear me , there’s nothing wrong with that..
But personally, I don’t pay much attention to benchmarks.
A model that scores higher on a benchmark isn’t necessarily better at what matters to us: understanding businesses, analyzing financials, valuing companies, and finding risks.
So I spent hours testing Fable 5.2 against Opus 5.5 on real companies, using my own Investing Skills.
And the results weren’t what I expected.

