Deep ultraviolet (DUV) phototransistors are key integral of optoelectronics bearing a wide spectrum of applications in flame sensor, military detector, oil spill detection, biological sensor, and artificial intelligence fields. In order to further improve the responsivity of UV photodetectors based on β-Ga₂O₃, in present work, high-performance β-Ga₂O₃ phototransistors with local back-gate structure were experimentally demonstrated. The phototransistor shows excellent DUV photoelectrical performance with a high responsivity of 1.01 × 10⁷ A/W, a high external quantum efficiency of 5.02 × 10⁹%, a sensitive detectivity of 2.98 × 10¹⁵ Jones, and a fast rise time of 0.2 s under 250 nm illumination. Besides, first-principles calculations reveal the decent stability of β-Ga₂O₃ nanosheet against oxidation and humidity without significant performance degradations. Additionally, the hexagonal boron nitride (h-BN)/β-Ga₂O₃ phototransistor can behave as a photonic synapse with ultralow power consumption of ~ 9.6 fJ per spike, which shows its potential for neuromorphic computing tasks such as facial recognition. This β-Ga₂O₃ phototransistor will provide a perspective for the next generation optoelectrical systems.