Issue Nº 01 — October 15, 2026

@readmodelmag
MODEL

Follow MODEL on X @readmodelmag

Best Tools · Coding

Tested

Parsimony

SummaryA benchmark that scores how many coding units AI agent patches add.

Filed October 13, 2026. Price, as published by the maker: Free, MIT. Python 3.12+. Analyzer has no dependencies..

Open the maker’s site (opens in a new tab)

Parsimony measures net coding units (AST elements) on published agent patches, not lines. Site: parsimony.lasu.dev. Code: github.com/La5u/Parsimony. Offline scripts/check.sh runs without network.

MODEL verified the README loop→sum example (net −10) and a same-task bloated vs tight compare. Limits (“smaller isn’t always better”) are the author’s. Full note in Issue Nº 02.