If tokens weren't an issue, which model would you choose, and why? Opus 5.5 versus Fable 5.1.

I'm wondering what model I should use for my client work and which one is smarter. I simply want to use the better, smarter model. For me, tokens are not an issue, so I'm trying to determine if anybody has run tests to see which one is better and what they found out. If you look at the data that Anthropic published, it seems like Opus 5.5 is a smarter model than Fable 5.1. Can you confirm this? Do your findings show the same?