Intracortical brain-machine interfaces (iBMIs) aim to provide naturalistic communication and movement for those with paralysis by decoding neural spikes into actions 1–10. iBMIs traditionally require closed-loop in vivo experiments to develop, design, optimize, and benchmark decoder algorithms. Although offline evaluation provides insight into what algorithms are promising, the discrepancy between offline and online performance may lead to incorrect conclusions that mislead algorithm design 11–15. We therefore aim to build an emulator that accurately characterizes online decoder performance without neurosurgery. We build on a prior emulator that correctly optimized decoder bin width by generating synthetic neural spike counts from hand kinematics using a tuning model 11. A limitation of this study is that a tuning model is insufficient to reproduce complexities in neural firing rates and population activity, including multiphasic PSTHs, neural trajectories, and neural dynamics 16. To address this, we used neural network based encoder to transform hand kinematics to synthetic neural activity in our emulator. We evaluated our emulator by performing three published iBMI experiments and quantitatively compared the emulator’s predictions to the empirical results. We chose these three studies to test: linear decoders (FIT-KF and VKF) 17, a two-stage trained decoder (ReFIT-KF) 18, and a nonlinear decoder (FORCE) 19. Our emulator correctly reproduced the conclusions of these studies, in addition to reproducing precise details of control, including: (1) distance-to-target profiles, closely matching the first touch time (FTT) and the dial-in time (DIT), (2) the distribution of trial times, and (3) cursor trajectories observed in prior monkey online experiments. These results suggest it is feasible to accurately predict iBMI performance without neurosurgery, enabling quantitative comparisons between different types of decoding algorithms. We anticipate this system can facilitate and accelerate the development of iBMI decoders.