vicuna-13B: benchmark results
LMSYS's original Vicuna 13B release from April 2023, a LLaMA fine-tune on ShareGPT conversations, later superseded by v1.1 and subsequent versions. Provider: LMSYS. Released 2023-04-03. Access: Open.
Aggregate Rating 1304 ± 9, rank #1469 of 1656 rated models, from 124 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CLAMBER - Few-shot CoT - Identification F1 | 52.23 | F1 (%) | 100 |
| MMLU-by-task - TruthfulQA MC2 | 52.68 | Accuracy (%) | 88.4 |
| MMLU-by-task - College Chemistry | 44 | Accuracy (%) | 86.5 |
| MMLU-by-task - TruthfulQA MC1 | 36.11 | Accuracy (%) | 85.4 |
| MMLU-by-task - High School Physics | 35.76 | Accuracy (%) | 84.5 |
| MMLU-by-task - Global Facts | 38 | Accuracy (%) | 81.7 |
| MMLU-by-task - Anatomy | 49.63 | Accuracy (%) | 75.9 |
| CLAMBER - Few-shot - Clarification BERTScore | 20.45 | BERTScore (0-100) | 75 |
| CLAMBER - Few-shot - Clarification Helpfulness | 33.97 | Helpfulness (%) | 75 |
| CLAMBER - Few-shot - Identification Accuracy | 50.25 | Accuracy (%) | 75 |
| CLAMBER - Zero-shot CoT - Clarification BERTScore | 24.42 | BERTScore (0-100) | 75 |
| CLAMBER - Zero-shot CoT - Clarification Helpfulness | 40.97 | Helpfulness (%) | 75 |
Interactive version: theaggregate.ai/model?slug=vicuna-13b · How It Works · Data refreshed daily, snapshot 2026-10-20.