Summarization quality depends on the question, of course mathematical or programming requests will benefit from reasoning models, but for some casual researches consisting of a few web-pages summarization is done ok even by mistral:7b, but as for the Detailed Report it would be better done by a reasoning model.