local/krea-2-turbo-no-filter Benchmark — 54/100 Overall, 192 Prompts, Every Image Shown

ImageBench V1 —

192 evaluations across 6 categories

Benchmark V1 verdicts are produced by VLM judges and can contain mistakes. Treat PASS/FAIL labels as machine-assisted assessments, and inspect the images yourself. Learn more about the methodology.

Generation Details

Source-backed model context, size, cost, and request settings for this run.

MakerKrea

FamilyKrea

Model size12B12B params

Costself-hostedfree

Latency71.5s median

Run targetlocal/krea-2-turbo-no-filter

Effective requestmodel: krea-2-turbo-no-filter · size: 1024x1024 · seed: 42

SourcesModel cardWeights & checkpoint

All 192 generations

Every challenge, grouped by category and subcategory. Solid-bordered tiles with a check passed the VLM judge; dashed tiles with a cross failed. Images are placeholders until generated.

Text Rendering

52%

Typography Style33%

Writing accuracy58%

Spatial Reasoning

53%

Attributes Binding56%

Compositionality56%

Counting67%

Negation44%

Relative Position67%

Scale & Proportions56%

Human realism

62%

Faces & Expressions58%

Full Body50%

Hands42%

Multi-Subject50%

Professional Studio

46%

Camera & Lighting42%

Color Precision25%

Photorealism33%

Graphical design

61%

Data Visualisation100%

Layout & Design56%

Style Diversity33%

Truthfulness

47%

Photorealism0%

Physics & Reflections42%

World Knowledge33%