Llama 3 8B Base: benchmark results
Provider: Meta. Released 2024-04-18. Access: Open.
Aggregate Rating 1407 ± 27, rank #5742 of 10916 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 66.49 | Accuracy (%) (5-shot) | 87.4 |
| Open LLM Leaderboard v1 - GSM8K | 45.34 | Accuracy (%) (5-shot) | 59.4 |
| Open LLM Leaderboard v1 - WinoGrande | 77.11 | Accuracy (%) (5-shot) | 53.3 |
| TheoremQA | 20.1 | Score (%) | 48.5 |
| Open LLM Leaderboard v1 - HellaSwag | 82.02 | Normalized accuracy (%) (10-shot) | 48.4 |
| Open LLM Leaderboard v1 - ARC Challenge | 59.22 | Normalized accuracy (%) (25-shot) | 42.4 |
| EvalPlus (HumanEval+ & MBPP+) | 40.5 | Pass@1 avg (%) | 33.9 |
| Science Leaderboard | 35.8 | Average Accuracy (%) | 27.3 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 43.95 | MC2 (%) (0-shot) | 25.3 |
Interactive version: theaggregate.ai/model?slug=llama-3-8b-base · How It Works · Data refreshed daily, snapshot 2026-11-08.