Use 'dd' to copy (in this case 3) blocks of data from the giant file to the output:
dd if=giant.json count=3 of=output.json
trucate could also work.
For JSON jq is a great tool:
Here we are catting json and selecting entries with messages length > 0
cat ~/giant.json | jq -cn --stream 'fromstream(1|truncate_stream(inputs))' | jq 'select((.messages | length) > 0)
