I think in fairness to anthropic they are winning in llms right? Since 3.7 they have been better than any other lab.
Arguably since 3.5, at least for coding and tool calling
Arguably since 3.5, at least for coding and tool calling