Llama 3 8B Instruct V0.8: benchmark results
MaziyarPanahi's Axolotl fine-tune of Meta Llama 3 8B Instruct, an iterative refinement built on his earlier v0.4 model. Provider: MaziyarPanahi. Released 2024-05-01. Access: Open.
Aggregate Rating 1465 ± 17, rank #908 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 52.13 | Normalized accuracy (%) | 92.2 |
| Open Korean LLM - Ko-IFEval | 46.3 | Strict accuracy, prompt/instruction mean (%) | 90.9 |
| Open Korean LLM - Ko-EQ-Bench | 39.59 | EQ-Bench score | 90.6 |
| Open LLM Leaderboard - IFEval | 75.28 | Score | 90.6 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 53.98 | Exact match, flexible extract (%) | 87.9 |
| Open Korean LLM Leaderboard | 43.34 | Average Score (%) | 87.8 |
| Open Ko-LLM Leaderboard | 43.34 | Average (%) | 87.7 |
| Open Portuguese LLM - ENEM | 71.52 | Accuracy (%) | 80.6 |
| Open Portuguese LLM - OAB Exams | 51.21 | Accuracy (%) | 79.4 |
| Open Portuguese LLM - BLUEX | 59.25 | Accuracy (%) | 76.1 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.25 | Normalized accuracy (%) | 75.9 |
| Open Korean LLM - KorNAT-Harmless | 64.58 | Normalized accuracy (%) | 75 |
Interactive version: theaggregate.ai/model?slug=llama-3-8b-instruct-v0-8 · How It Works · Data refreshed daily, snapshot 2026-10-20.