vicuna-13B-v1.5-16k: benchmark results
LMSYS's 13B chat assistant, a 16K-context version of Vicuna v1.5 using linear RoPE scaling, fine-tuned from Llama 2 on about 125,000 ShareGPT conversations. Provider: LMSYS. Released 2023-08-01. Access: Open.
Aggregate Rating 1323 ± 23, rank #1417 of 1635 rated models, from 84 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Language - Content Summarization - English (CompassBench 2404) | 30.4 | Score (%) | 75 |
| OpenCompass Language - Translation (CompassBench 2404) | 15.4 | Score (%) | 68.3 |
| OpenCompass Language - Content Summarization - Chinese (CompassBench 2404) | 25.1 | Score (%) | 66.7 |
| OpenCompass Reasoning - Abductive - English (CompassBench 2404) | 68 | Score (%) | 63.8 |
| OpenCompass Agent - CIBench - Chinese (CompassBench 2404) | 38.6 | Score (%) | 63.3 |
| OpenCompass LLM - Knowledge - English (CompassBench 2404) | 70.7 | Score (%) | 60 |
| OpenCompass LLM - Reasoning - English (CompassBench 2404) | 28.8 | Score (%) | 60 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 51.96 | MC2 (%) (0-shot) | 53.9 |
| OpenCompass Reasoning - Inductive - English (CompassBench 2404) | 27.7 | Score (%) | 53.3 |
| OpenCompass Agent - CIBench - English (CompassBench 2404) | 39 | Score (%) | 50 |
| OpenCompass LLM - Agent (CompassBench 2404) | 43.2 | Score (%) | 50 |
| OpenCompass LLM - Agent - Chinese (CompassBench 2404) | 44 | Score (%) | 50 |
Interactive version: theaggregate.ai/model?slug=vicuna-13b-v1-5-16k · How It Works · Data refreshed daily, snapshot 2026-10-15.