Llama-VARCO-8B-Instruct: benchmark results
NC Research's Korean-specialist fine-tune of Llama 3.1 8B, continually pretrained and preference-tuned for Korean while keeping English (September 2024). Provider: Meta. Released 2024-09-12. Access: Open.
Aggregate Rating 1552 ± 49, rank #1674 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LogicKor - Writing | 9.71 | Score (0-10) | 96.8 |
| LogicKor - Grammar | 8.21 | Score (0-10) | 95 |
| LogicKor - Multi-Turn | 9.07 | Score (0-10) | 95 |
| LogicKor | 8.83 | Score (0-10) | 93.3 |
| LogicKor - Reasoning | 8.71 | Score (0-10) | 90.8 |
| LogicKor - Single-Turn | 8.59 | Score (0-10) | 89.7 |
| LogicKor - Math | 7.78 | Score (0-10) | 89.5 |
| LogicKor - Understanding | 9.5 | Score (0-10) | 87 |
| LogicKor - Coding | 9.07 | Score (0-10) | 80.7 |
| Open LLM Leaderboard - GPQA | 29.7 | Score | 52.5 |
| Open LLM Leaderboard - BBH | 50.23 | Score | 49.3 |
| Open LLM Leaderboard - MATH Level 5 | 10.65 | Score | 49.2 |
Interactive version: theaggregate.ai/model?slug=llama-varco-8b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.