fal/bytedance/seedream-v4 Benchmark — 59/100 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.

MakerByteDance

FamilySeedream

Model sizenot disclosedundisclosed

Cost$0.03 / imagelow

Latency14.1s median

Run targetfal/bytedance/seedream-v4

Effective requestmodel: seedream-v4 · size: 1024x1024 · seed: 42

SourcesByteDance model docsByteDance pricing

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

74%

Typography Style67%

Writing accuracy67%

Spatial Reasoning

58%

Attributes Binding44%

Compositionality33%

Counting67%

Negation67%

Relative Position50%

Scale & Proportions56%

Human realism

71%

Faces & Expressions67%

Full Body58%

Hands67%

Multi-Subject67%

Professional Studio

72%

Camera & Lighting58%

Color Precision58%

Photorealism100%

Graphical design

51%

Data Visualisation33%

Layout & Design56%

Style Diversity25%

Truthfulness

49%

Photorealism0%

Physics & Reflections33%

World Knowledge50%