In view of the lag of universal time (UT1) products, which cannot meet the real-time requirement of users, it is proposed to fuse the ultra-rapid variation of length of day (ΔLOD) sequence with the UT1 sequence observed by very long baseline interferometry (VLBI) through the Vondrak algorithm. In the fusion process, combined with the Markov chain Monte Carlo simulation method, the weight ratio selection of UT1 and ΔLOD is studied to obtain the optimal real-time UT1 product. It fills the lack of ultra-rapid UT1 products at home and abroad, which provides data support for UT1 prediction. By using this method, the UT1 sequence of national time service center and ultra-rapid (ΔLOD) sequence of iGMAS, UT1 sequence of national time service center and ultra-rapid (ΔLOD) sequence of IGS, UT1 sequence of IVS and ultra-rapid(ΔLOD) sequence of IGS are fused respectively, and the fused data are compared with IERS C04 UT1 sequence. Results show that real-time UT1 products can be achieved after ultra-rapid sequence fusion, the accuracy of UT1 reaches 45. 91 μs and 87. 40 μs, respectively. They are better than the daily forecast products of IERS.