Vicuna-7B: benchmark results
LMSYS's original Vicuna 7B chat assistant from April 2023, fine-tuned from LLaMA on user-shared ShareGPT conversations, the first release in the Vicuna family. Provider: LMSYS. Released 2023-04-05. Access: Open.
Aggregate Rating 1247 ± 15, rank #1542 of 1656 rated models, from 40 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LitBench - Citation Recommendation (Quantum Physics) | 13.1 | Accuracy (%) at picking the paper actually cited, out of 11 | 33.3 |
| LitBench - Citation Recommendation (Robotics) | 11.75 | Accuracy (%) at picking the paper actually cited, out of 11 | 33.3 |
| T-Eval | 44.8 | Overall Score (%) | 25 |
| ClinicBench | 34.61 | Average Score | 23.8 |
| AlpacaEval 1.0 | 64.41 | Win Rate (%) | 17.8 |
| LitBench - Citation Recommendation (Quantitative Biology) | 12.5 | Accuracy (%) at picking the paper actually cited, out of 11 | 16.7 |
| BenchBench | 18.85 | Aggregate Score (%) | 15.4 |
| JustEval - Depth | 4.04 | Score (1-5) | 13.3 |
| JustEval - Factuality | 4.33 | Score (1-5) | 13.3 |
| JustEval - Helpfulness | 4.43 | Score (1-5) | 13.3 |
| ENAMEL | 6.1 | Greedy eff@1 (%) (imported) | 10.3 |
| JustEval - Clarity | 4.85 | Score (1-5) | 10 |
Interactive version: theaggregate.ai/model?slug=vicuna-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.