MotionLCM is a real-time controllable motion generation model that achieves high-quality text-to-motion and precise motion control results in approximately 30 milliseconds. The model is based on a motion latent consistency model (MotionLCM), which extends controllable motion generation to real-time applications. By employing one-step or few-step inference, MotionLCM improves the runtime efficiency of motion generation. A motion ControlNet is incorporated within the latent space to enable explicit control signals, such as pelvis trajectory, to control the generation process. This allows for direct control of motion generation, similar to controlling other latent-free diffusion models. The model's effectiveness is demonstrated through experimental results showing a good balance of generation quality, controlling capability, and real-time efficiency. MotionLCM is compared with other state-of-the-art methods on the HumanML3D dataset, achieving superior performance in terms of inference speed and motion quality. The model's key contributions include the introduction of latent consistency distillation for motion generation and the development of a motion ControlNet for real-time motion control. The model's performance is validated through quantitative and qualitative experiments, demonstrating its effectiveness in generating high-quality, controllable motion.MotionLCM is a real-time controllable motion generation model that achieves high-quality text-to-motion and precise motion control results in approximately 30 milliseconds. The model is based on a motion latent consistency model (MotionLCM), which extends controllable motion generation to real-time applications. By employing one-step or few-step inference, MotionLCM improves the runtime efficiency of motion generation. A motion ControlNet is incorporated within the latent space to enable explicit control signals, such as pelvis trajectory, to control the generation process. This allows for direct control of motion generation, similar to controlling other latent-free diffusion models. The model's effectiveness is demonstrated through experimental results showing a good balance of generation quality, controlling capability, and real-time efficiency. MotionLCM is compared with other state-of-the-art methods on the HumanML3D dataset, achieving superior performance in terms of inference speed and motion quality. The model's key contributions include the introduction of latent consistency distillation for motion generation and the development of a motion ControlNet for real-time motion control. The model's performance is validated through quantitative and qualitative experiments, demonstrating its effectiveness in generating high-quality, controllable motion.