In this Chapter we discuss employing GPs as pricing surrogates, providing a fast prediction of contracts’ values as a function of their parameters. Thus, we consider the use of GPs to map a state \(\mathbf {x}\) (which might include underlying asset price, but also deterministic contract parameters like its strike) into a option price \(P(\mathbf {x})\) . The different sections cover learning derivative valuations within a probabilistic framework (Sect. 4.1); ingredients of GP surrogates for option pricing (Sect. 4.2); using GPs to learn option Greeks and other sensitivities (Sect. 4.3); imposing no-arbitrage constraints in GP models (Sect. 4.4); and applications to portfolio modeling and credit value adjustment computation (Sect. 4.5). The Chapter includes an R-based notebook that illustrates the learning of option prices and Deltas within the Black-Scholes and Heston models.