PECC: leaderboard
Programming Error Correction and Comprehension benchmark scoring models on understanding and fixing buggy code from natural-language problem descriptions.
Metric: Pass@3. Source: huggingface.co. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3 Haiku | 27.67 | #803 |
| 2 | GPT-3.5 Turbo | 23.75 | #865 |
| 3 | Mixtral 8x7B Instruct | 8.35 | #1170 |
| 4 | Phi-3 Mini 128K Instruct | 7.18 | #1298 |
| 5 | WizardLM-2-7B | 3.72 | #1395 |
| 6 | Llama 3 8B Instruct | 3.1 | #1129 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=pecc · How It Works · Data refreshed daily, snapshot 2026-10-20.