Primate Labs, the benchmarking pioneers, published Geekbench AI 1.0 on Thursday. The program, which is now available for Android, Linux, MacOS, and Windows, applies Geekbench ideas to machine learning, deep learning, and other AI tasks in an effort to standardize performance ratings across platforms. It is the successor to Geekbench ML (machine learning), which was launched in 2021 and is now at version 0.6.
"[I]n recent years, companies have coalesced around using the term 'AI' in these kinds of workloads (and in their related marketing)," Primate Labs notes of the moniker shift. "To ensure that everyone, from engineers to performance enthusiasts, understands what this benchmark does and how it works, we felt it was time for an update."
Earlier this week, OpenAI, the creator of ChatGPT, published a new edition of its AI model benchmark. SWE-bench Verified is a "human-validated" product that assesses models' efficacy in addressing "real-world issues."