Geekbench AI 1.0 is available to download as an app and supports Android, iOS, Linux, macOS, and Windows.
Geekbench AI can measure Single-Precision, Half-Precision, and Quantized scores
Photo Credit: Geekbench
Geekbench AI 1.0 platform, which can measure the performance of artificial intelligence (AI) capabilities of a device, was launched on Thursday. Developed by Primate Labs, the app is a benchmarking suite dedicated to measuring and evaluating the overall AI-centric performance of devices. It is available to download for free across all major platforms. The AI tool can run several tests on the CPU, GPU, and neural processing unit (NPU) to generate a score for the device. Developers also have the option to choose the right AI framework and models to test workloads.
Announcing the app, the company said in a blog post, “Geekbench AI is a benchmarking suite with a testing methodology for machine learning, deep learning, and AI-centric workloads, all with the same cross-platform utility and real-world workload reflection that our benchmarks are well-known for.”
Geekbench AI iOS app (tested iPhone 15 Pro)
The company highlighted that the app runs ten different AI workloads automatically, where each test requires three different data types. This comprehensive testing helps users get a better assessment of the on-device AI performance. The app is available for Android, iOS, Linux, macOS, and Windows and can evaluate smartphones, tablets, laptops, desktops, and similar devices.
Interestingly, the preview release of the app was named Geekbench ML, but it was renamed as the company noticed that OEMs have started using the word AI to describe these workloads. Further, to handle the complexity of determining AI performance, the benchmarking app considers the workloads, hardware, and AI framework of the device.
Essentially, the Geekbench AI app tests the device for speed and accuracy, as this helps determine if the device makes any trade-offs between performance and efficiency. Other such metrics include datasets, frameworks, runtime, computer vision, natural language processing (NLP), and more.
Primate Labs have also created an ML Benchmarks leaderboard where users can check the CPU, GPU, and NPU performance of different devices, and know the top-performing devices. The minimum software requirements to run the app are as follows. Android 12 or later,
Catch the latest from the Consumer Electronics Show on Gadgets 360, at our CES 2026 hub.