Blueprint-BenchBrowse 296

Blueprint-Bench

Can the model extract and reason over information in images or documents?

Version not specifiedExact reported variant
Vision8 ranked models9 reported values2 reportsHigher is better

Top rankings

One row per model, using its best reported score across effort settings.

8
RankModelBest scoreBest reported setting
1GPT-5.5OpenAI36.2%Reported configurationAnthropic report2 values · 2 reportsJun 9, 2026 · source
2Gemini 3.5 FlashGoogle33.6%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source
3Gemini 3.1 Pro PreviewGoogle26.5%Reported configurationAnthropic report1 value · 1 reportJun 9, 2026 · source
4Gemini 3.1 ProGoogle26.5%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source
5Claude Opus 4.7Anthropic24.5%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source
6Claude Opus 4.8Anthropic14.5%Reported configurationAnthropic report1 value · 1 reportJun 9, 2026 · source
7Claude Sonnet 4.6Anthropic6.7%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source
8Gemini 3 FlashGoogle0.0%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source

Effort curve

Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.

0
No cost-linked effort sweep for this version.
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.

Normalized score.

Method / source