In this paper, two models of OPAQ air quality prediction system based on artificial neural network algorithm are used to analyze the accuracy of O₃ prediction. Taking the OPAQ system of Jiangsu Suqian Environmental Center as an example, the results show that the correlation coefficients R of RMSE model for predicting the current day and the next three days are above 0.78 and the relative errors are below 25%. The prediction accuracy of RMSE model for the current day and the 24hour, 48hour and 72hour excellentgood days is high, which are 88.8%, 87.2%, 86.3% and 84.7% respectively. The accuracy of RMSE model for predicting the lightheavy pollution is low. The correlation coefficients R of the extreme optimal model (SI model) for predicting the current day and the next three days are above 0.76, and the relative errors are below 32%. The accuracy of predicting the level of mildmoderate pollution in the next 24 hours and 48 hours is above 60%. In summary, the extreme optimal model (SI model) of OPAQ system is more suitable as a prediction tool when the concentration of O₃ is high in summer.