local/sefi-image-2b-turbo Benchmark — 34/100 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.

MakerSefi

FamilySefi Image

Model size2B2B params

Costself-hostedfree

Latency3.6s median

Run targetlocal/sefi-image-2b-turbo

Effective requestmodel: sefi-image-2b-turbo · 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

30%

Typography Style33%

Writing accuracy17%

Spatial Reasoning

40%

Attributes Binding44%

Compositionality33%

Counting33%

Negation22%

Relative Position58%

Scale & Proportions33%

Human realism

49%

Faces & Expressions50%

Full Body42%

Hands50%

Multi-Subject50%

Professional Studio

52%

Camera & Lighting33%

Color Precision50%

Photorealism33%

Graphical design

47%

Data Visualisation33%

Layout & Design22%

Style Diversity25%

Truthfulness

30%

Photorealism0%

Physics & Reflections17%

World Knowledge42%