It depends. If you're using Python with numpy>=2.0.0 (and macOS>=14) then you should benefit greatly from Apple's Accelerate implementation of BLAS/LAPACK routines which are behind most linear algebra operations. I'm not aware of any serious public benchmarks, though.