WizardLM-13B-V1.2: benchmark results
The WizardLM team's WizardLM-13B V1.2, a full-weight instruction-following fine-tune of Llama 2 13B, part of the WizardLM family alongside WizardCoder and WizardMath. Provider: Microsoft. Released 2023-07-25. Access: Open.
Aggregate Rating 1331 ± 11, rank #1399 of 1635 rated models, from 149 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - High School Mathematics | 32.22 | Accuracy (%) | 89.1 |
| MMLU-by-task - Public Relations | 67.27 | Accuracy (%) | 87.5 |
| MMLU-by-task - Security Studies | 66.94 | Accuracy (%) | 86.8 |
| MMLU-by-task - Medical Genetics | 64 | Accuracy (%) | 86.3 |
| MMLU-by-task - International Law | 75.21 | Accuracy (%) | 84.2 |
| MMLU-by-task - Jurisprudence | 73.15 | Accuracy (%) | 84 |
| MMLU-by-task - High School Computer Science | 60 | Accuracy (%) | 83.8 |
| MMLU-by-task - Professional Medicine | 56.25 | Accuracy (%) | 82.5 |
| MMLU-by-task - Anatomy | 51.11 | Accuracy (%) | 82.4 |
| MMLU-by-task - Clinical Knowledge | 61.13 | Accuracy (%) | 82.4 |
| MMLU-by-task - HellaSwag | 62.95 | Accuracy (%) | 80.6 |
| MMLU-by-task - High School Chemistry | 43.84 | Accuracy (%) | 80.6 |
Interactive version: theaggregate.ai/model?slug=wizardlm-13b-v1-2 · How It Works · Data refreshed daily, snapshot 2026-11-05.