MOPAC could bring the quantum precision you need to study large molecules or periodic systems. A good trade-off between speed and accuracy is achieved through a minimal basis and parameterization against experimental data, with parameters available for most elements.
Like DFTB, the semi-empirical MOPAC code uses the nearest neighbor and minimal basis set approximations, making it fast and linear scaling. MOPAC has been parameterized against an enormous set of thoroughly examined experimental data, in a huge, commendable effort by Dr. Jimmy Stewart. The latest parameterization set (PM7) is also the most accurate. Since AMS2019, the MOPAC and DFTB modules are now bundled together and only one license is needed to use both methods.
To accelerate your research, use MOPAC for quick pre-optimization, prescreening of a large number of molecules or crystals, or to get results on really large molecules which are out of reach for DFT or ab initio methods.