Llama 3 70B Instruct DPO V0.2: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1576 ± 17, rank #750 of 2928 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 88.25 | Accuracy (%) (5-shot) | 100 |
| Open LLM Leaderboard v1 - MMLU | 80.41 | Accuracy (%) (5-shot) | 99.7 |
| Open Chinese LLM - GSM8K | 71.42 | Accuracy (%) | 98.2 |
| Open Chinese LLM - CMMLU | 72.29 | Accuracy (%) | 95.3 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.53 | Normalized accuracy (%) (25-shot) | 92.4 |
| Open Chinese LLM - ARC Challenge | 60.41 | Accuracy (%) | 88.4 |
| Open Chinese LLM Leaderboard | 66.97 | Average Score (%) | 86.4 |
| Open LLM Leaderboard v1 - WinoGrande | 82.79 | Accuracy (%) (5-shot) | 85.6 |
| Open Chinese LLM - C-Eval Semantic | 86.94 | Accuracy (%) | 82.8 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 63.57 | MC2 (%) (0-shot) | 80.9 |
| Open LLM Leaderboard v1 - HellaSwag | 86.22 | Normalized accuracy (%) (10-shot) | 79.6 |
| Open Chinese LLM - HellaSwag | 63.4 | Accuracy (%) | 73 |
Interactive version: theaggregate.ai/model?slug=llama-3-70b-instruct-dpo-v0-2 · How It Works · Data refreshed daily, snapshot 2026-09-23.