The OpenCL Working Group today released the OpenCL 3.1.2 specifications. This update finalizes the cl_khr_command_buffer extension, alongside various bug fixes and clarifications to the standard. The primary benefit of the finalized command buffer feature is its ability to let applications record a sequence of compute commands once and then replay them repeatedly. This significantly reduces CPU overhead and provides opportunities for OpenCL drivers to optimize across the entire recorded sequence. Workloads with repeated execution patterns, such as those commonly found in AI inference, are expected to see performance improvements. The complete OpenCL 3.1.2 specifications are now available in the OpenCL Registry.