EXAONE-3.5-2.4B-Instruct: benchmark results
LG AI Research's smallest EXAONE 3.5 instruct model (2.4B), Korean/English bilingual with a 32K context (December 2024). Provider: LG AI. Released 2024-12-09. Access: Open.
Aggregate Rating 1346 ± 23, rank #1391 of 1656 rated models, from 43 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - IFEval | 79.5 | Score | 96.8 |
| Open LLM Leaderboard - MATH Level 5 | 36.78 | Score | 87.2 |
| SLM-RAG Arena | 1567.2 | Elo Rating | 81.2 |
| Horangi 4 - Ko-HalluLens (Nonexistent Entities) | 79 | Refusal rate (%) | 63 |
| Mobile-MMLU | 53.7 | Overall Accuracy (%) | 47.6 |
| Open LLM Leaderboard - MMLU-Pro | 32.8 | Score | 45.4 |
| Horangi 4 - KorQuAD 1.0 | 77.58 | F1 (x100) | 38.5 |
| Open LLM Leaderboard - BBH | 40.92 | Score | 25.6 |
| Horangi 4 - Ko-HLE | 9 | Accuracy (%) | 23.6 |
| Open LLM Leaderboard - MuSR | 36.61 | Score | 21.6 |
| Open LLM Leaderboard - GPQA | 26.59 | Score | 20.3 |
| Horangi 4 - Ko-MT-Bench | 80.75 | Judge rating (1-10, x10) | 13.5 |
Interactive version: theaggregate.ai/model?slug=exaone-3-5-2-4b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.