granite-8B-code-base: benchmark results
IBM's Apache-2.0 8B code base model from the May 2024 Granite Code family, pretrained on ~4T tokens across 116 programming languages. Provider: IBM. Released 2024-05-06. Access: Open.
Aggregate Rating 1323 ± 48, rank #8278 of 10916 rated models, from 12 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Portuguese NLU - HAREM | 49.33 | Micro F1 excluding MISC (%) | 75.5 |
| ShaderMatch | 13.96 | Clone Match Rate (%) | 61.9 |
| EuroEval Portuguese NLU - MultiWikiQA PT | 68.63 | Character F1 (%) | 49.2 |
| EuroEval Portuguese NLU | 48.47 | NLU Average Score (%) | 45.8 |
| EuroEval Portuguese NLU - SST-2 PT | 72.75 | Matthews correlation coefficient (%) | 40.3 |
| EuroEval Portuguese | 34.71 | Average Score (%) | 34.2 |
| DuckDB-NSQL | 45.3 | Execution Accuracy (%) | 30.2 |
| EuroEval Portuguese NLU - ScaLA PT | 3.17 | Matthews correlation coefficient (%) | 30.1 |
| EuroEval Portuguese Knowledge | 12.94 | Knowledge Average Score (%) | 22.8 |
| EuroEval Portuguese Common Sense Reasoning | 1.46 | Common Sense Reasoning Average Score (%) | 21 |
| EuroEval Danish Knowledge | 5.25 | Knowledge Average Score (%) | 16.8 |
| EuroEval Danish Knowledge - Danish Citizen Tests | 5.25 | MCC (x100) | 13.4 |
Interactive version: theaggregate.ai/model?slug=granite-8b-code-base · How It Works · Data refreshed daily, snapshot 2026-11-08.