PAIR-Bench - Targeted Repair: leaderboard
Targeted repair success on PAIR-Bench (2026): the share of the hidden tests in the hinted failure scenario that pass after each repair turn, averaged over turns and initially failing programs, in percent. Higher is better.
Metric: Targeted repair success (%; share of the hidden tests in the hinted failure scenario that pass after the repair turn, averaged over turns and initially failing instances; higher is better). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V3.2 | 71.59 | #406 |
| 2 | Gemini 2.5 Flash Lite | 59 | #958 |
| 3 | GPT-4o Mini | 56.14 | #1623 |
| 4 | Qwen 3 Coder 30B A3B Instruct | 55.89 | #972 |
| 5 | Llama 3.3 70B Instruct | 53.5 | #1357 |
| 6 | Ministral 3 14B | 49.71 | #1845 |
| 7 | Mistral Small 3.2 | 49.31 | #1591 |
| 8 | Gemma 3 27B | 42.56 | #1631 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=pair-bench-targeted-repair · How It Works · Data refreshed daily, snapshot 2026-11-13.