The massive success of blockchains has significantly catalyzed interest in the extensive deployment of practical asynchronous Byzantine fault-tolerant (BFT) consensus protocols across wide area networks. However, existing asynchronous consensus protocols suffer from high communication cost and low throughput, which can’t meet the high-performance requirements of blockchain platforms. We identify two key factors that hinder the efficiency of asynchronous BFT protocols: (1) Reliable broadcast protocols incur high communication costs; (2) The broadcast and agreement phases are executed strictly sequentially, resulting in wasted bandwidth resources. To address above issues, we propose the lightweight provable broadcast and a novel asynchronous BFT protocol called TP-BFT. TP-BFT implements a lightweight broadcast followed by a recovery mechanism and supports parallel broadcast, agreement, and recovery phases. This approach offers two key advantages: (1) it mitigates the impact of low-bandwidth nodes on the efficiency of the broadcast phase; (2) it fully utilizes the bandwidth during the agreement phase to enhance the protocol’s throughput. We implement TP-BFT and conduct experiments with HoneyBadger and Dumbo. The results demonstrate that TP-BFT improves throughput by 106% and reduces latency by 64% compared to Dumbo.