Perplexity / Claude doesn’t win all the benchmarks or sport the biggest parameter count or context space but I still favor it. It definitely has the most natural, “human-like” conversational style. If I feel his answer is inadequate I provide a little more context in my prompt or would use “deep search” (don’t think I’ve had to use it yet).
I do find that if I submit a simple prompt I tend to get a simple answer, whereas a prompt with “completeness of query with the subject well fleshed out” elicits a more comprehensive answer.