How do the commercial embedding models compare against each other? Eg Cohere vs OpenAI small vs OpenAI large etc? I have troubles navigating this space as ther

How do the commercial embedding models compare against each other? Eg Cohere vs OpenAI small vs OpenAI large etc? I have troubles navigating this space as there’s so much choice, and I don’t know exactly how to “benchmark” an embedding model for my use cases.