Geekbench launches cross-platform AI benchmark

Primate Labs just launched Geekbench AI 1.0, bringing its cross-platform benchmarking playbook to the Wild West of local AI performance.

Now available on Windows, macOS, Linux, and Android, the tool applies standardized scoring to machine learning, deep learning, and everyday AI workloads across completely different hardware ecosystems.

Goodbye ML, hello AI

If the tool sounds familiar, that is because it is the official successor to Geekbench ML, which had been parked at version 0.6 since its 2021 debut.

Primate Labs admitted the rebrand is largely pragmatic. Tech companies and marketing teams have rallied around the term "AI" in recent years, so the team updated the name to make sure everyone from casual hardware enthusiasts to system engineers immediately understands what the benchmark actually measures.

Why it matters

Chipmakers and PC brands love throwing around proprietary NPU specs and theoretical TOPS numbers that tell you next to nothing about real-world performance. Geekbench AI gives consumers and developers a neutral, cross-platform baseline to see how hardware actually handles AI tasks.

We think this is a smart move that injects much-needed transparency into the market. However, AI workloads evolve far faster than traditional CPU benchmarks. The real test is whether synthetic scores can genuinely reflect real-world experiences like local text generation and on-device image processing.