Parsimony - Open-source benchmark for AI code bloat
Show HN: Parsimony – an open-source benchmark for how much code AI models add
Parsimony is an open-source benchmark that measures how much code AI coding models add, not just solve rate or cost. It counts coding units like statements and calls, subtracting deleted code to compute net units added. Using patches from SWE-bench Verified, it ranks models by mean net units added on solved tasks, revealing which AI agents write the least code. The tool helps developers choose efficient models that keep codebases maintainable. You can even add your own benchmark data via a simple JSON file, all processed in your browser.
Existing benchmarks measure solve rate and cost, but don't punish excess code, which makes a codebase harder to work with in the long term.