Source-Test Repair - Failure log Compilable: leaderboard

Source-test co-evolution repair on 500 Java cases from 15 repositories; Java17/Maven3.9.9, temperature0.6. Native fractions retain denominator500. Executable means repaired tests pass, not semantic equivalence to developer intent; compilable measures build validity. Context and pipeline configurations are separate. Failure log; Compilable.

Metric: Compilable fraction. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 3 models tracked.

Top models

#ModelScoreOverall rank
1GPT-5.4 (low) (TestRepair Run: context=Failure%20log; pipeline=baseline)0.45
2Qwen3-Coder-480B-Instruct (TestRepair Run: context=Failure%20log; pipeline=baseline)0.36
3DeepSeek-V3.2 (Non-reasoning) (TestRepair Run: context=Failure%20log; pipeline=baseline)0.3

Interactive version: theaggregate.ai/benchmark?slug=source-test-repair-failure-log-compilable · How It Works · Data refreshed daily, snapshot 2026-11-05.