jq is pretty fast in my experience. But there have been cases where I've wanted it to be faster (dealing with a 90GB JSON file).

The main weakness seems to be streaming use cases (not having the whole file in memory at once). These are supported, but the syntax is quite awkward.