JSON is hard to read in bulk and cannot be opened as a table without flattening. TSV avoids the quoting problems commas cause in CSV. JSON is at home in APIs and data exchanged between programs; TSV is the usual choice for tabular data whose values contain commas.
One JSON can produce several TSV files; when it does, they are offered together as a ZIP download.
A list of records becomes a table with one row per record. Nested keys turn into columns named like address.city, list items into columns like tags.0, and a key that only some records have leaves empty cells in the others. If the top level holds several lists of records, each becomes its own table, saved as a separate file; a document that is a single object becomes one row.