SWE-bench Evolving: Opus 4.6 Leads at 79.2%, But Harder Variants Expose Limits

VALS AI·low signal

Opus 4.6 (Thinking) leads SWE-bench Verified at 79.2%, but SWE-Bench Pro (~23%) and SWE-EVO (19-21%) reveal significant gaps. Current coding agents excel at well-defined single-repo issues but struggle with cross-language, cross-repo, evolutionary tasks — the reality of production engineering.