VOLTBench - Mean Length Accuracy (5 Sections): leaderboard
Metric: Mean length accuracy (%): 100 x max(0, 1 - |mean output length - target length| / target length) over the five sampled generations per instruction, English simple-difficulty instructions, Story (unstructured) and Code Function (structured) chapter-based tasks, 5-section target; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 | 92.3 |
| 2 | Llama 3.1 8B Instruct | 60.6 |
| 3 | GPT-4o Mini | 59 |
| 4 | DeepSeek V3 | 56.2 |
| 5 | Qwen 2.5 7B Instruct | 49.5 |
| 6 | Claude 3.5 Sonnet | 43.7 |
| 7 | Qwen 2.5 1.5B Instruct | 24.9 |
Interactive version: theaggregate.ai/benchmark?slug=voltbench-mean-length-accuracy-5-sections · How It Works · Data refreshed daily, snapshot 2026-10-07.