@samhenrigold The thing I did notice going from M1 to M3 was anything that used the Neural Engine was significantly faster on M3