VOLTBench - Mean Length Accuracy (20 Sections): leaderboard

Length adherence on VOLTBench (2026), a long-form generation benchmark of chapter-based story and code instructions, at the 20-section target: five generations are sampled per instruction and their mean length is scored as 100 x max(0, 1 - |mean - target| / target). Percent; higher is better.

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, 20-section target; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1LongWriter-8B92.8
2GPT-4o Mini51.7#1667
3DeepSeek R146.3#482
4DeepSeek V329.2#729
5Qwen 2.5 7B Instruct22.9#2723
6Qwen 2.5 1.5B Instruct10.4#7470
7Falcon3-Mamba-7B-Instruct8.8
8Llama 3.1 8B Instruct6.7#4600
9Claude 3.5 Sonnet4.5#752

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=voltbench-mean-length-accuracy-20-sections · How It Works · Data refreshed daily, snapshot 2026-11-21.