Arena AI - Document Leaderboard: leaderboard
Crowdsourced Elo-rated leaderboard for document understanding: PDF analysis, long-content reasoning, and document Q&A through blind pairwise battles.
Metric: Elo. Source: arena.ai. Source: benchmarklist.com. 44 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (High) | 1516 | #12 (Claude Opus 5) |
| 2 | Claude Fable 5.1 (Max) | 1513 | #9 (Claude Fable 5.1) |
| 3 | Claude Opus 4.6 | 1507 | #63 |
| 4 | Claude Opus 4.6 (High) | 1507 | #63 (Claude Opus 4.6) |
| 5 | Claude Fable 5 | 1504.25 | #16 |
| 6 | Claude Fable 5 (High) | 1496 | #16 (Claude Fable 5) |
| 7 | Claude Opus 4.7 | 1495 | #49 |
| 8 | Claude Opus 4.7 (High) | 1495 | #49 (Claude Opus 4.7) |
| 9 | GPT-5.5 | 1486 | #31 |
| 10 | GPT-5.5 (High) | 1484 | #31 (GPT-5.5) |
| 11 | GPT-5.6 Sol (xHigh) | 1483 | #19 (GPT-5.6 Sol) |
| 12 | Claude Sonnet 4.6 | 1482 | #90 |
| 13 | Claude Opus 4.8 (High) | 1475 | #37 (Claude Opus 4.8) |
| 14 | GPT-5.6 Terra (xHigh) | 1472 | #41 (GPT-5.6 Terra) |
| 15 | GPT-5.4 | 1471 | #79 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=arena-ai-document-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.