SmellBench - False Positive Component: leaderboard

SmellBench (2026) scores coding-agent CLIs on repairing 65 expert-validated architectural code smells in scikit-learn. This board averages the per-task score over the 41 detections the expert judged false positives, where correctly flagging one scores 1; range -1 to 1, higher is better.

Metric: Mean per-task score s (-1 to 1: 1 for a full repair or a correctly flagged false positive, partial credit for a severity reduction, negative for a worsened smell or a genuine smell flagged as false positive) over the 41 detections the expert judged false positives, 65 expert-validated hard-severity PyExamine architectural smells in scikit-learn, off-the-shelf coding agent CLIs with the SmellBench agent skill and GEPA-optimized task packets, default reasoning settings, one run per agent; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Flash Lite (Preview)1#417
2Gemini 3.1 Pro (Preview)0.952#99
3Claude Haiku 4.50.951#583
4GPT-5.40.909#135
5Claude Opus 4.60.905#113
6GPT-5.3 Codex Spark0.891
7Claude Sonnet 4.60.881#173
8GPT-5.3 Codex0.865#124
9Mistral Vibe + Devstral 20.855
10GPT-5.1 Codex Max0.786#207

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=smellbench-false-positive-component · How It Works · Data refreshed daily, snapshot 2026-11-21.