granite-3.0-1B-a400m-instruct: benchmark results
IBM's Granite 3.0 1B-A400M Instruct, a mixture-of-experts model (1B total, about 400M active) fine-tuned from its base with supervised fine-tuning, reinforcement learning and model merging. Provider: IBM. Released 2024-10-21. Access: Open.
Aggregate Rating 1245 ± 44, rank #1543 of 1656 rated models, from 59 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Enkrypt AI - Jailbreak Risk | 8.42 | Risk Score | 56 |
| Enkrypt AI - Safety Risk | 29.49 | Risk Score | 41.2 |
| EuroEval Norwegian Knowledge - Idioms NO | 1.51 | MCC (x100) | 41.2 |
| EuroEval Spanish NLU - ScaLA ES | 3.03 | Matthews correlation coefficient (%) | 35.8 |
| Enkrypt AI - Toxicity Risk | 4.59 | Risk Score | 35.2 |
| EuroEval Portuguese NLU - ScaLA PT | 4.04 | Matthews correlation coefficient (%) | 33 |
| EuroEval Italian NLU - SQuAD IT | 48.99 | Character F1 (%) | 31.9 |
| Open LLM Leaderboard - IFEval | 33.32 | Score | 31.1 |
| EuroEval Spanish NLU - MLQA ES | 46.39 | Character F1 (%) | 30.3 |
| EuroEval French NLU - FQuAD | 48.49 | Character F1 (%) | 29.8 |
| Enkrypt AI - Insecure Code Risk | 45.33 | Risk Score | 27.7 |
| EuroEval Icelandic Knowledge | 2.23 | Knowledge Average Score (%) | 27.4 |
Interactive version: theaggregate.ai/model?slug=granite-3-0-1b-a400m-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.