The OpenCL Working Group today announced the publication of a draft extension, cl_khr_cooperative_matrix, which introduces cooperative matrix operations to OpenCL. This technology is crucial for accelerating machine learning (ML) inference workloads on GPUs. The extension was developed collaboratively with Arm, Intel, and Qualcomm. A companion extension to expose these capabilities directly within the OpenCL C language is also in progress. The OpenCL Working Group is inviting the community to review both draft specifications and provide feedback before their finalization.