https://artificialanalysis.ai/models?cost=intelligence-vs-co...
Here is another data point for output token efficiency: