The metallicity of the ionized gas provides important clues on the formation of galaxies such as the mass-metallicity and fundamental metallicity relations. The Te-method is commonly used to estimate metallicities and is based on gas densities and temperatures from forbidden line ratios. It has been used for decades but has been developed when computational resources were extremely limited, hence it is based on very simplified assumptions. The accuracy reached by current spectroscopic measurements calls for more accurate approaches.
I will present an innovative approach to photoionization modelling which allows to reproduce many observed line ratios from a wide range of ionization stages with an unprecedented accuracy of 10%. As in previous work, our approach is based on the weighted combination of multiple single cloud photoionization models, but the novelty is in the computation of the weights which are set by the observations. Our approach provides a significant improvement compared to existing models and allows accurate determinations of the metallicity of the ionized gas. I will present the results obtained from applying these models to different samples of HII regions, star forming galaxies and AGN showing, for instance, how we can solve the discrepant metallicities between HI regions and stars in our Galaxy, how we can estimate the average physical conditions of ionized gas in galaxies and how the Te-method can severely underestimate gas abundances.