- Flexible parallelism – you can now use more than one thread per partition to speed up individual queries
- New vector functions (VECTOR_SORT and more) that support financial Value-at-Risk calculations and other applications
- The SET statement is now supported for session variable assignment
- BACKUP WITH SPLIT PARTITIONS works with unlimited storage databases
- Computed column expressions can be matched by the query optimizer, including for JSON expressions
- Hash-group-by spilling is on by default
- SECRET() function prevents passwords from appearing in logs and query plan output
For the full feature list, see the release notes.