Transition metal oxides have attracted intense interest owing to their abundant physical and chemical properties. The controlled preparation of large-area, high-quality two-dimensional crystals is essential for revealing their inherent properties and realizing high-performance devices. However, fabricating two-dimensional (2D) transition metal oxides using a general approach still presents substantial challenges. Herein, we successfully achieve highly crystalline nickel oxide (NiO) flakes with a thickness as thin as 3.3 nm through the salt-assisted vapor–liquid–solid (VLS) growth method, which demonstrated exceptional stability under ambient conditions. To explore the great potential of the NiO crystal in this work, an artificial synapse based on the NiO-flake resistive switching (RS) layer is investigated. Short-term and long-term synaptic behaviors are obtained with external stimuli. The artificial synaptic performance provides the foundation of the neuromorphic application, including handwriting number recognition based on artificial neuron network (ANN) and the virtually unsupervised learning capability based on generative adversarial network (GAN). This pioneering work not only paves new paths for the synthesis of 2D oxides in the future but also demonstrates the substantial potential of oxides in the field of neuromorphic computing.