DocScope: leaderboard
Evaluating whether Multimodal Large Language Models can produce trustworthy, verifiable reasoning over long, visually rich documents requires evaluation beyond end-to-end answer accuracy. We introduce DocScope, a benchmark that formulates long-document QA as a structured reasoning trajectory prediction problem: given a complete PDF document and a question, the model outputs evidence pages, support
Metric: ACC All (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 78.9 | #58 |
| 2 | Claude Opus 4.7 | 76 | #49 |
| 3 | Claude Sonnet 4.6 | 70.5 | #90 |
| 4 | Gemini 3.1 Flash Lite | 68.1 | #191 |
| 5 | Qwen 3.6 Plus | 66.8 | #120 |
| 6 | Qwen 3.5 27B | 66.4 | #187 |
| 7 | Qwen 3.5 397B A17B | 63.8 | #146 |
| 8 | Qwen 3.5 122B A10B | 63.6 | #173 |
| 9 | GPT-5.4 | 62.7 | #79 |
| 10 | Gemma 4 31B | 59.4 | #200 |
| 11 | Qwen 3 VL 235B A22B | 51.4 | #238 |
| 12 | Ministral 3 14B | 41.5 | #650 |
| 13 | Ministral 3 8B | 36.2 | #687 |
| 14 | Gemma 4 26B A4B | 34.4 | #358 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=docscope · How It Works · Data refreshed daily, snapshot 2026-10-20.