WSLG supports faux GPU passthrough/sharing with the host, e.g. for CUDA workloads or graphics acceleration from within the WSL environment.