Using sequence-to-sequence models for abstractive text sum-marization is generally plagued by three problems: inability to deal with out-of-vocabulary words,repetition in summaries and time-consuming in training.The paper proposes a hierarchical hybrid neural network archi-tecture for Chinese text summarization.Three mechanisms,hierarchical attention mechanism,pointer mechanism and coverage mechanism,are integrated into the architecture to improve the performance of summa-rization.The proposed model is applied to Chinese news headline gener-ation.The experimental results suggest that the model outperforms the baseline in ROUGE scores and the three mechanisms can improve the quality of summaries.