openchat-3.6-8B-20240522 (2024-05-22): benchmark results
OpenChat's openchat-3.6-8b-20240522, a Llama 3 8B fine-tune using OpenChat's mixed-quality-data training approach. Provider: OpenChat. Released 2024-05-22. Access: Open.
Aggregate Rating 1432 ± 13, rank #1090 of 1656 rated models, from 61 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 71.8 | Accuracy (%) (5-shot) | 96.7 |
| Open Korean LLM - Ko-IFEval | 43.85 | Strict accuracy, prompt/instruction mean (%) | 88.3 |
| Open LLM Leaderboard v1 - MMLU | 66.56 | Accuracy (%) (5-shot) | 87.6 |
| Open Korean LLM Leaderboard | 42.55 | Average Score (%) | 86.6 |
| Open Ko-LLM Leaderboard | 42.55 | Average (%) | 86.5 |
| Open PL LLM - DYK (generative, 5-shot) | 66.16 | Binary F1 (%) | 78 |
| Open PL LLM - PSC (generative, 5-shot) | 96.69 | Binary F1 (%) | 78 |
| Open PL LLM - PPC (multiple choice, 5-shot) | 77.6 | Accuracy (%) | 77.9 |
| Open PL LLM - DYK (multiple choice, 5-shot) | 65.38 | Binary F1 (%) | 77.8 |
| Open PL LLM - PolQA Reranking (multiple choice, 5-shot) | 79.11 | Accuracy (%) | 77.7 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 42.61 | Exact match, flexible extract (%) | 77.6 |
| Open PL LLM - PoQuAD Open Book (generative, 5-shot) | 68.65 | Levenshtein Similarity (%) | 77 |
Interactive version: theaggregate.ai/model?slug=openchat-3-6-8b-20240522-2024-05-22 · How It Works · Data refreshed daily, snapshot 2026-10-20.