Chartography
Can the model extract and reason over information in images or documents?
Top rankings
One row per model, using its best reported score across effort settings.
| Rank | Model | Best score | Best reported setting |
|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 89.0% | max effortAnthropic report1 value · 1 reportSep 22, 2026 · source |
| 2 | Claude Fable 5.1Anthropic | 88.4% | Reported configurationAnthropic report1 value · 1 reportSep 22, 2026 · source |
| 3 | Claude Opus 5Anthropic | 83.4% | Reported configurationAnthropic report1 value · 1 reportSep 22, 2026 · source |
| 4 | DeepSeek V4.1 FlashDeepSeek | 78.9% | Reported configurationDeepSeek report1 value · 1 reportSep 10, 2026 · source |
Effort curve
Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.
Definition and comparison boundarypublic methodology
Can the model extract and reason over information in images or documents?
Visual perception and multimodal reasoning under the reported protocol. Higher is better. The value is the percentage reported in this lab's table.
Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.
DeepSeek's official V4.1-Flash release table; exact source label Chartography (w/tools) retained. Anthropic's release table reports the exact with-tools scores; it does not print the task revision, harness, or model effort for the comparison cells beyond its general Opus 5.5 adaptive-Max note.
Method / source