What buyers ask about inference endpoints

Short answers, no provider named, no number from the private scorecards. Each page links to the method behind the grade.

Why does the same model give different answers on different providers?

Because the model is only the weights. Each provider adds a serving stack: precision, inference engine, batching, sampling defaults, the context window it honors, and changes after listing.

fp8 vs fp4 vs bf16: what the precision field tells you and what it doesn't

It is the provider's declaration of the number format. It tells you the format. It does not tell you what the endpoint returns.