VOLTBench - Structured Content Accuracy (500 Sections): leaderboard

Structured content accuracy on VOLTBench (2026) at the 500-section target: the share of the 500 required chapters of a generated code library that pass execution-based verification, averaged over five sampled generations. Percent; higher is better.

Metric: Structured content accuracy (%): share of the 500 required chapters of the Code Function task that pass execution-based verification, averaged over the five sampled generations per instruction, English simple-difficulty instructions, Story (unstructured) and Code Function (structured) chapter-based tasks; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 2.5 7B Instruct33.7#2723
2GPT-4o Mini32.6#1667
3Llama 3.1 8B Instruct28.2#4600
4LongWriter-8B26.8
5Falcon3-Mamba-7B-Instruct20.8
6DeepSeek V39.4#729
7DeepSeek R13.4#482
8Qwen 2.5 1.5B Instruct1.8#7470
9Claude 3.5 Sonnet0.6#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-structured-content-accuracy-500-sections · How It Works · Data refreshed daily, snapshot 2026-11-21.