DocScope - Page Localization: leaderboard

DocScope (2026) evaluates verifiable reasoning over long, visually rich PDF documents: for each human-annotated question of its 730-question test split the model reads the whole document and returns evidence pages, evidence regions, supporting facts and an answer. This board scores the evidence pages by micro-averaged F1 against the gold pages, in percent; higher is better.

Metric: Page localization F1 (%): micro-averaged F1 of the predicted evidence pages against the gold pages by exact page match over the answerable questions of the 730-question DocScope test split (human-annotated questions on long, visually rich PDF documents; the model receives the whole document and returns evidence pages, evidence regions, supporting facts and an answer), each stage scored only on the gold pages the model retrieved; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)87.6#99
2Claude Opus 4.786#85
3Claude Sonnet 4.684#173
4Gemini 3.1 Flash Lite83.8#371
5Qwen 3.5 27B82.1#375
6Qwen 3.6 Plus78.9#238
7GPT-5.478.4#135
8Gemma 4 31B76.9#382
9Qwen 3.5 397B A17B74.1#264
10Qwen 3 VL 235B A22B72.1#485
11Qwen 3 VL 32B68.7#622
12Ministral 3 14B58.4#1870
13Gemma 4 26B A4B56.4#590
14Qwen 3.5 122B A10B55#346
15Ministral 3 8B53#2131

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

Interactive version: theaggregate.ai/benchmark?slug=docscope-page-localization · How It Works · Data refreshed daily, snapshot 2026-11-21.