CodeLlama-34B: benchmark results
Provider: Meta.
Aggregate Rating 1356 ± 35, rank #7545 of 10916 rated models, from 16 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BizBench | 62.5 | Mean accuracy (%) over the eight BizBench tasks (FinKnow, Fo | 71.4 |
| BizBench - CodeFinQA | 52.4 | Accuracy (%; 844 FinQA questions over report text and tables | 66.7 |
| BizBench - CodeTAT-QA | 81.4 | Accuracy (%; 392 TAT-QA table questions answered by generati | 66.7 |
| BizBench - FinCode | 17.4 | Accuracy (%; 121 CFA and CPA exam questions answered by gene | 66.7 |
| BizBench - FormulaEval | 70 | Accuracy (%; 50 financial-formula functions to complete from | 66.7 |
| BizBench - SEC-Num | 67 | Accuracy (%; 2,000 SEC 10-K and 10-Q filing snippets, extrac | 66.7 |
| Big Code Memorization - HumanEval-ET pass@1 | 41.46 | HumanEval-ET pass@1 (%) | 58.8 |
| Big Code Memorization - HumanEval pass@1 | 48.17 | HumanEval pass@1 (%) | 55.9 |
| BizBench - ConvFinQA Extract | 88.1 | Accuracy (%; 916 ConvFinQA questions answerable by extractin | 53.3 |
| BizBench - TAT-QA Extract | 83.9 | Accuracy (%; 248 TAT-QA questions answerable by extracting a | 53.3 |
| Big Code Memorization - HumanEval pass@50 | 34.78 | HumanEval pass@50 (%) | 52.9 |
| Big Code Memorization - HumanEval-ET pass@50 | 29.06 | HumanEval-ET pass@50 (%) | 52.9 |
Interactive version: theaggregate.ai/model?slug=codellama-34b · How It Works · Data refreshed daily, snapshot 2026-11-08.