Tests indicate that the Apple M5 Max is outpacing AMD and establishing a new performance benchmark.

      The M5 Max has just outperformed every processor on Geekbench—yes, all of them.

      The initial unofficial Geekbench 6 scores for the 16-inch MacBook Pro featuring the M5 Max have been released, and Apple is not merely leading the benchmark competition, but is decisively exceeding it.

      The 18-core CPU of the M5 Max achieved a multi-core performance score of 29,233, surpassing the Mac Studio’s M3 Ultra chip, which has a 32-core CPU and an average score of 27,726. That’s correct; an 18-core chip has outperformed a 32-core one. This indicates a roughly 5% advantage over the M3 Ultra and is as much as 15% faster than the previous year’s M4 Max.

      But that's not all. In terms of single-core performance, the M5 Max received a score of 4,268, marking the highest single-core result of any consumer PC processor evaluated on Geekbench, including AMD's Ryzen 9 series. Apple has claimed that its super cores are the fastest CPU cores in the world, and the benchmarks seem to support this assertion.

      What contributes to the speed of the M5 Max?

      The M5 Max is constructed on a new Fusion Architecture that combines two dies into a single chip, integrating a powerful CPU, scalable GPU, and enhanced unified memory bandwidth.

      The 18-core CPU comprises six super cores designed for maximal speed and 12 new performance cores optimized for efficient multithreaded tasks. Additionally, the M5 Max can support up to 128GB of unified memory with bandwidth up to 614GB/s. This innovative Fusion architecture has enabled Apple to enhance performance beyond previous limits, delivering remarkable results with its M5 Max processors.

      How does the GPU perform?

      The GPU results offer a more complex picture. The M5 Max’s 40-core GPU achieved scores of 218,772 and 232,718 in two different Metal tests, falling about 5% to 10% behind the M3 Ultra’s average score of 245,053, but exceeding by over 20% the performance of the MacBook Pro that features an M4 Max.

      However, comparisons with Nvidia’s and AMD’s GPUs are not yet possible, as the relevant test results have not been released. Furthermore, GPU performance can vary significantly based on the software used or the games played. Thus, we can’t assess how the GPU of the M4 Max will perform until it is available for shipment.

      Notably, the Nvidia RTX 5090 has consistently scored nearly double that of the M3 Ultra GPU in tests such as Geekbench 6 OpenCL and 3DMark, suggesting that the M5 Max may perform in a similar range.

      In summary, the M5 Max provides up to 15% improved CPU performance and 20% enhanced GPU performance compared to its M4 Max predecessor. If you have been considering upgrading your MacBook Pro, these performance metrics could help sway your decision. Pre-orders are open, with availability beginning on March 11.

      Rachit is an experienced tech journalist with over seven years in the consumer technology sector.

      Google is preparing its Gemini AI app for Mac, introducing Desktop Intelligence.

      The forthcoming app could allow Gemini to interface directly with Mac apps for more tailored responses.

      Google is enhancing its AI offerings on desktop with a dedicated Gemini app for Mac currently undergoing testing, potentially including a significant new feature called Desktop Intelligence. According to a Bloomberg report, Google has begun private testing of an early version of the Gemini Mac app with a select group of users. This initiative is part of a larger effort to more directly compete with desktop AI applications like ChatGPT and Claude, both of which already provide native Mac experiences.

      The present version is still in development and lacks several planned features, but testing indicates Google is aiming for a release soon. Currently, Mac users mainly use Gemini through the web, so having a dedicated app could greatly improve user experience.

      In a different development, a startup is offering $800 a day to work as an AI bully.

      This presents an opportunity to monetize frustrations with chatbots.

      If you've ever expressed frustration at an AI chatbot for not remembering previous information, you might be ready for this job. A startup named Memvid is offering $800 for just one day of work, where your sole responsibility is to bully AI. Yes, it's as unconventional as it sounds. The job entails spending eight hours communicating with chatbots, asking them to recall information, observing their failures, and then posing additional questions. Your task is to document each frustrating occurrence and every detail or context that the AI forgets.

      Additionally, a Microsoft PowerToys update is introducing three significant enhancements for your Windows PC.

      PowerToys 0.98 has launched with an improved Keyboard Manager, a pinnable Command Palette Dock, and a more reliable CursorWrap for those utilizing multiple monitors.

      Microsoft has rolled out PowerToys v0