This study presents a new volcanic sulfur dioxide (SO₂) layer height retrieval algorithm for the Ozone Monitoring Instrument (OMI). We generated a large spectral dataset with a radiative transfer model and used it to train neural networks to predict SO₂ height from OMI radiance data. The algorithm is fast and takes less than 10 min for a single orbit. Retrievals were tested on four eruption cases, and results had reasonable agreement (within 2 km) with other retrievals and previous studies.