AI TrackerFrontier desk · AEST
Back to news
normalaitech

Show HN: Parsimony – an open-source benchmark for how much code AI models add

Hacker News · 14 Oct 2026, 05:50 AEST

Parsimony is an open-source benchmark designed to measure the amount of code added by AI models. It focuses on coding units rather than lines of code to assess the impact on long-term project maintainability.

Key facts

Read aloud

Parsimony is an open-source benchmark designed to measure the amount of code added by A I models. It focuses on coding units rather than lines of code to assess the impact on long-term project maintainability. Existing benchmarks measure solve rate and token output/cost, but don't punish excess code, which makes a codebase harder to work with in the long term. Method: it takes the patches that benchmarks already publish DeepSWE, swe bench Verified, swe bench Live and counts 'coding units': syntax elements like statements, calls, names and literals. I didn't use lines of code, so cramming code isn't rewarded. Deleted code is subtracted, so it measures net units. Any feedback or model contributions are welcome, especially public patch exports for newer models.

Sources in this cluster