It’s not easy being one of Silicon Valley’s favorite benchmarks. 

SWE-Bench (pronounced “swee bench”) launched in November 2024 to evaluate an AI model’s coding skill, using more than 2,000 real-world programming problems pulled from the public GitHub repositories of 12 different Python-based projects. 

In the months since then, it’s quickly become one of the most popular tests in AI. A SWE-Bench score has become a mainstay of major model releases from OpenAI, Anthropic, and Google—and outside of foundation models, the fine-tuners at AI firms are in constant competition to see who can rise above the pack. The top of the leaderboard is a pileup between three different fine tunings of Anthropic’s Claude Sonnet model and Amazon’s Q developer agent. Auto Code Rover—one of the Claude modifications—nabbed the number two spot in November, and was acquired just three months later.