local/flux-2-klein-4b Benchmark — 52/100 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.

MakerBlack Forest Labs

FamilyFLUX.2

Model size4B4B params

Costself-hostedfree

Latency4.5s median

Run targetlocal/flux-2-klein-4b

Effective requestmodel: flux-2-klein-4b · 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

36%

Typography Style33%

Writing accuracy25%

Spatial Reasoning

54%

Attributes Binding33%

Compositionality44%

Counting56%

Negation44%

Relative Position50%

Scale & Proportions44%

Human realism

63%

Faces & Expressions50%

Full Body67%

Hands67%

Multi-Subject50%

Professional Studio

66%

Camera & Lighting50%

Color Precision58%

Photorealism67%

Graphical design

36%

Data Visualisation67%

Layout & Design33%

Style Diversity17%

Truthfulness

36%

Photorealism67%

Physics & Reflections33%

World Knowledge42%