3D printing is a promising method for the generation of complex‐shaped optical components. However, the printing efficiency, form accuracy, and surface smoothness are still challenging, especially for the printing of optical components. In this study, by synergizing the volumetric projection stereolithography, meniscus equilibrium effect, and iterative learning scheme, a low‐cost volumetric 3D printing method is developed for the ultra‐fast and high‐precision fabrication of miniature lenses with sub‐nanometric roughness. By including the covered liquid film as a part of the printed lens in the iterative learning, a low form error at the micrometer scale is achieved for the printed lens without any prior knowledge of the 3D printing process. As a demonstration, a sphere lens at the millimeter scale is directly printed in 2 s, achieving a peak‐to‐valley profile error of less than 5 µm, a sub‐nanometric roughness of rms = 0.614 nm, and an imaging resolution of 203.2 lp mm−1. It suggests that besides fast prototyping, the developed 3D printing strategy is suitable for the massive production of precise lenses.