GLM-5.2
ZhipuRank #4 of 11 · 29 PRs · 32/91 golden bugs found
40.2
F1 · +2.7 vs avg
Recall
35.2%-0.1 vs avg
Precision
47.0%+2.5 vs avg
Cost / PR
$0.880
Cost / bug found
$0.79
Confidence
95% bootstrap interval on recall (2000 resamples over the 29 PRs):24.7–47.1points. This measures sampling variance from which PRs are in the set.
Judge run-to-run variance not yet measured for this model.
Neither measures model run-to-run variance — re-running the review agent itself, not just the judge. One pass per entry.
Run configuration
HarnesskodusAccess pathapiExecution modereplayReasoningvendor-defaultJudgeclaude-haiku-4-5
Severity mix
What the model called its own findings — not recall by severity, goldens aren't severity-tagged.
Low 12Medium 33High 19Critical 4
Category mix
Only bug/performance/security are consistent across models — the rest is free text, bucketed as other.
Bug66Performance3Security3Other11