Benchmarking giant Primate Labs released Geekbench AI 1.0 on Thursday. Currently available for Android, Linux, MacOS, and Windows, the app applies Geekbench principles to machine learning, deep learning, and other AI workloads, aiming to standardize performance evaluation across platforms. It's the successor to Geekbench ML (machine learning), which was announced in 2021 and is currently at version 0.6.
“[I]”In recent years, companies have become more united in using the term 'AI' for these types of workloads (and their associated marketing),” Primate Labs said about the name change. “We felt it was time to update it to help everyone, from engineers to performance enthusiasts, understand what this benchmark does and how it works.”
Earlier this week, ChatGPT developer OpenAI announced a new version of its proprietary AI model benchmark: SWE-bench Verified, a “human-validated” product that uses human validation to determine the effectiveness of models at solving “real-world problems.”