The open circuit voltage (OCV) is a chemistry-dependent curve that characterizes the steady-state behaviour of a lithium-ion battery cell (LIB), namely the link between the voltage accross the cell and its state of charge (SOC), when no load is connected. It is a fundamental ingredient to estimate the SOC of LIBs with any technique, ranging from a simple look-up table to a Kalman filter based on an equivalent circuit model or an electrochemical model. However, an accurate determination of such a steady-state curve for a commercial LIB requires extensive standalone experimental campaigns that are time-consuming. This work, compares two different methods, Karhunen-Loéve transform (or principal component analysis) and Gaussian process regressions, to model the zero-current behaviour of 18650 LIBs (Sony VTC6 and Samsung 30Q, 3000mAh) considering OCV-SOC charge and discharge experimental curves at C-rates between C/50 and C/5. The zero-current extrapolation given by these methods can reduce the experimental time up to 85% when compared to an average OCV characterized at C/100.