Llama 3 8B Instruct V0.9: benchmark results
MaziyarPanahi's iterative general-chat fine-tune of Meta Llama 3 8B Instruct, built on top of his v0.8 model. Provider: MaziyarPanahi. Released 2024-05-30. Access: Open.
Aggregate Rating 1471 ± 15, rank #884 of 1656 rated models, from 35 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 53.53 | Normalized accuracy (%) | 94.5 |
| Open Korean LLM - Ko-IFEval | 50.8 | Strict accuracy, prompt/instruction mean (%) | 94.1 |
| Open LLM Leaderboard - IFEval | 76.3 | Score | 91.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.35 | Normalized accuracy (%) (25-shot) | 91.7 |
| Open LLM Leaderboard v1 - MMLU | 68.1 | Accuracy (%) (5-shot) | 90.6 |
| Open Korean LLM Leaderboard | 44.57 | Average Score (%) | 90.5 |
| Open Ko-LLM Leaderboard | 44.57 | Average (%) | 90.3 |
| Open LLM Leaderboard v1 - HellaSwag | 88.17 | Normalized accuracy (%) (10-shot) | 90 |
| Open Korean LLM - Ko-EQ-Bench | 37.51 | EQ-Bench score | 88.2 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 53.3 | Exact match, flexible extract (%) | 86.9 |
| Open Korean LLM - Ko-GPQA-Diamond | 26.26 | Normalized accuracy (%) | 85.5 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 64.67 | MC2 (%) (0-shot) | 82.6 |
Interactive version: theaggregate.ai/model?slug=llama-3-8b-instruct-v0-9 · How It Works · Data refreshed daily, snapshot 2026-10-20.