CategoryAgentic
MeasureMicro-average percentage of bugs resolved with a repository regression gate
Tasks4,068 reference bugs across 100 repositories in 22 languages
DifficultyRepository-wide bug discovery and repair
Tests whether coding agents can find and fix multiple bugs across a repository without issue descriptions, while preserving existing behavior.
No model in the catalog has a published SWE-sweep score.The benchmark is defined by SWE-sweep: Can Agents Autonomously Find and Fix Bugs?, but the catalog carries no value for it yet. An absent value is shown as absent here rather than as a zero.
About SWE-sweep
What the benchmark measures and how its value is published, from the source’s own record.
What is SWE-sweep?
SWE-sweep (SWE-sweep: Can Agents Autonomously Find and Fix Bugs?) is an agentic benchmark published by SWE-sweep: Can Agents Autonomously Find and Fix Bugs. Tests whether coding agents can find and fix multiple bugs across a repository without issue descriptions, while preserving existing behavior. Its evaluation set is described by the source as 4,068 reference bugs across 100 repositories in 22 languages. The source rates its difficulty as repository-wide bug discovery and repair.
How is SWE-sweep scored?
Each model's result is published by SWE-sweep: Can Agents Autonomously Find and Fix Bugs as micro-average percentage of bugs resolved with a repository regression gate. ModelScale prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 0 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
ModelScale