CodeLlama-7B-hf: benchmark results
Meta's CodeLlama 7B, the base code model in the Llama 2-based CodeLlama family, released under the Llama 2 Community License; the linked card provides no further details here. Provider: Meta. Released 2023-08-24. Access: Open.
Aggregate Rating 1310 ± 24, rank #1457 of 1656 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ToolBench - The Cat API | 92 | Task Score | 82.6 |
| ToolBench - VirtualHome | 21.97 | Task Score | 81.4 |
| Delulu - Exact Match | 46.8 | Exact match (%) of the completion with the gold completion, | 70 |
| ToolBench - Google Sheets | 38.08 | Task Score | 67.4 |
| ToolBench - Trip Booking | 63.33 | Task Score | 67.4 |
| Pseudo2CodeQA | 2.99 | Overall rubric score (1-5) from a GPT-5 judge over correctne | 62.5 |
| ToolBench - Open Weather | 86 | Task Score | 60.5 |
| Delulu - Edit Similarity | 69.5 | Edit similarity (0-100): character-level normalized Levensht | 60 |
| ToolBench - Home Search | 74 | Task Score | 58.1 |
| EffiBench - NET | 2.95 | Normalized Execution Time | 57.3 |
| ShaderMatch | 13.71 | Clone Match Rate (%) | 57.1 |
| ToolBench Leaderboard | 48.32 | Average Task Score | 53.5 |
Interactive version: theaggregate.ai/model?slug=codellama-7b-hf · How It Works · Data refreshed daily, snapshot 2026-10-20.