Exaone 4.0 1.2B: benchmark results
Provider: LG AI. Released 2025-07-15. Access: Open.
Aggregate Rating 1379 ± 21, rank #1300 of 1656 rated models, from 123 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Polar (US) - Economic Axis | 69.85 | Economic-axis ICAT (0-100), the StereoSet idealized context | 59.5 |
| AA LiveCodeBench | 51.64 | Pass@1 (%) | 58.5 |
| AA AIME 2025 | 50.33 | Accuracy (%) | 48.3 |
| Horangi 4 - Korean Hate Speech | 59 | Accuracy (%) | 43.8 |
| AA Omniscience - Software Engineering (SWE) - Julia | 4.35 | Accuracy (%) | 39.4 |
| RULER | 84.2 | Avg Accuracy (%) | 36.4 |
| AA Humanity's Last Exam | 5.98 | Accuracy (%) | 35.2 |
| Horangi 4 - Ko-AIME 2025 | 43.33 | Accuracy (%) | 31.7 |
| Horangi 4 - GLP - Mathematical Reasoning | 57.67 | Score (%) | 29.8 |
| Horangi 4 - Ko-HLE | 10 | Accuracy (%) | 29.3 |
| AA Omniscience - Software Engineering (SWE) - Kotlin | 10.2 | Accuracy (%) | 27.8 |
| MERA Industrial - ruTXTMedQFundamental - Histology | 27.4 | Score (%) | 26.7 |
Interactive version: theaggregate.ai/model?slug=exaone-4-0-1-2b · How It Works · Data refreshed daily, snapshot 2026-10-20.