Focused ultrasound featuring non-destructive and high sensitivity has attracted widespread attention in biomedical and industrial evaluation. However, most traditional focusing techniques focus on the design and improvement of single-point focusing, neglecting the need to carry more dimensions of multifocal beams. Here we propose an automatic multifocal beamforming method, which is implemented using a four-step phase metasurface. The metasurface composed of four-step phases improves the transmission efficiency of acoustic waves as a matching layer and enhances the focusing efficiency at the target focal position. The change in the number of focused beams does not affect the full width at half maximum (FWHM), revealing the flexibility of the arbitrary multifocal beamforming method. Phase-optimized hybrid lenses reduce the sidelobe amplitude, and excellent agreement is observed between the simulation and experiments for triple-focusing beamforming metasurface lenses. The particle trapping experiment further validates the profile of the triple-focusing beam. The proposed hybrid lens can achieve flexible focusing in three dimensions (3D) and arbitrary multipoint, which may have potential prospects for biomedical imaging, acoustic tweezers, and brain neural modulation.