Granite 3.3 8B Instruct: benchmark results
Provider: IBM. Released 2025-02-26. Access: Open.
Unified ELO 1447 ± 1, rank #957 of 1392 rated models, from 45 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BlueBench - Safety | 88.44 | Score (%) | 100 |
| LatamBoard - FLORES Bidirectional | 43.78 | Score (%) | 81.8 |
| LatamBoard - Spanish XNLI | 48.07 | Score (%) | 81.8 |
| LatamBoard - Translation Score | 45.03 | Score (%) | 81.8 |
| LatamBoard - Portuguese Score | 86.96 | Score (%) | 78.8 |
| BlueBench - RAG General | 51.92 | Score (%) | 76.5 |
| LatamBoard - Spanish COPA | 82.8 | Score (%) | 74.2 |
| LatamBoard - Spanish Score | 62.01 | Score (%) | 72.7 |
| LatamBoard - ASSIN2 RTE | 91.91 | Score (%) | 69.7 |
| HELM AIR-Bench | 76 | Refusal Rate (%) | 68.6 |
| LatamBoard - BLUEX | 57.86 | Score (%) | 66.7 |
| LatamBoard - Spanish OpenBookQA | 35.6 | Score (%) | 65.2 |
Interactive version: theaggregate.ai/model?slug=granite-3-3-8b-instruct · How It Works · Data refreshed daily, snapshot 2026-09-05.