Hi. When it comes to dbt it really depends on the data , the logic of the sql query or more precisely the data it self. you can build your own model to run for normalization in dbt BUT even so we are talking about 1 min less max for high amount of data. You can build a model that will only extract the fields you need. adding more threads also wont work since they are good mostly for ref() functions in dbt or in some cases for high number of sources for 1 model. simply put it is what it is, since we are talking about taking json data and extract it and the extraction is linear. or if you have faster CPU

Thank you for the explanation.