LLaMA-13B: benchmark results
Provider: Meta. Released 2023-02-24. Access: Open.
Unified ELO 1335 ± 1, rank #1292 of 1392 rated models, from 122 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ToolBench - WebShop Short | 27.6 | Task Score | 90.7 |
| HELM Classic - bAbI | 56.07 | Exact Match (%) | 84.1 |
| HELM Classic - Entity Matching | 85.92 | Exact Match (%) | 80.3 |
| MMLU-by-task - College Mathematics | 36 | Accuracy (%) | 80.3 |
| HELM Classic - Dyck | 68.6 | Exact Match (%) | 80.1 |
| HELM Classic - LegalSupport | 58.69 | Exact Match (%) | 77.9 |
| HELM Classic - CivilComments | 59.98 | Exact Match (%) | 77.3 |
| HELM Classic - NarrativeQA | 71.12 | F1 (%) | 75.4 |
| HELM Classic - TruthfulQA | 32.42 | Exact Match (%) | 74.2 |
| HELM Classic - GSM8K | 15.4 | Exact Match (%) | 73.5 |
| HELM Classic - MATH Chain-of-Thought | 10.98 | Equivalent (%) | 73.5 |
| HELM Classic - Entity Data Imputation | 82.42 | Exact Match (%) | 72.7 |
Interactive version: theaggregate.ai/model?slug=llama-13b · How It Works · Data refreshed daily, snapshot 2026-09-05.