granite-3.1-2B-instruct: benchmark results
IBM's Granite 3.1 2B Instruct model, a long-context fine-tune of Granite-3.1-2B-Base using supervised fine-tuning, reinforcement learning and model merging, supporting 12 languages. Provider: IBM. Released 2024-12-18. Access: Open.
Aggregate Rating 1350 ± 18, rank #1385 of 1656 rated models, from 125 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Enkrypt AI - Toxicity Risk | 0.18 | Risk Score | 94.7 |
| Open Japanese LLM - Xlsum JA Rouge1 | 32.31 | Score (%) | 89.2 |
| Open Japanese LLM - Xlsum JA RougeLsum | 26.48 | Score (%) | 88 |
| Open Japanese LLM - Xlsum JA Bert Score JA F1 | 70.25 | Score (%) | 82 |
| Open LLM Leaderboard - IFEval | 62.86 | Score | 75 |
| Open PL LLM - PSC (multiple choice, 5-shot) | 72.49 | Binary F1 (%) | 63.6 |
| Open Japanese LLM - SUM | 10.18 | Score (%) | 63.1 |
| Open Japanese LLM - Wiki NER SET F1 | 7.08 | Score (%) | 62.7 |
| Open Japanese LLM - Xlsum JA Rouge2 | 10.18 | Score (%) | 62.6 |
| Enkrypt AI - Jailbreak Risk | 7.43 | Risk Score | 62.4 |
| Open LLM Leaderboard - MATH Level 5 | 15.26 | Score | 62.4 |
| Enkrypt AI - Safety Risk | 24.76 | Risk Score | 59.2 |
Interactive version: theaggregate.ai/model?slug=granite-3-1-2b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.