Abstract: High-resolution intelligent grid forecast has become the main operational work and the development direction of future weather forecast since China Meteorological Administration launched the seamless fine grid weather forecast service in 2014.To date, with the rapid development of numerical model statistical post-processing technology, the quality of grid temperature forecast has been steadily improved.On the basis of summarizing the existing fine-scale testing and evaluation methods for grid temperature forecast, this paper systematically sorts out the latest progress of numerical model interpretation techniques, such as single model correction, multi-model integrated interpretation and big data artificial intelligence, in the field of objective correction of grid temperature forecast.The advantages and disadvantages of different methods in the field of grid temperature forecast are elucidated.In the context of the current rapid development of new information technologies such as artificial intelligence, big data, and cloud computing, grid temperature forecast needs to harness more cutting-edge research results from multiple sources, and fully utilize the inherent forecast information of numerical models for a smarter forecast in the future.