AlphaFold3 (aka AF3) is a newer version of AF2 using diffusion for predicting the 3D structure of proteins + nucleic acids + small molecules + post-translational modification. It is capable of multi-chains and therefore is also a cofolding or docking method.
ESMFold is an ML model that is a variation of AlphaFold2 that offers higher speed.
Folding is the problem of taking a sequence (string) and predicting the structure (3D coordinates). Also called structure prediction, folding usually applies to proteins, but RNA folding is also studied. Protein structures are useful for understand how a protein functions and providing a starting point for making changes to improve the function. Because experimental structure determination using cryo-EM or crystallography are slow, expensive, and sometimes impossible, computational structure prediction can be very valuble.