Ghost Palette

fal/ideogram/v4 — ImageBench V1

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.

MakerIdeogram
FamilyIdeogram
Model sizenot disclosedundisclosed
Cost$0.015 / MPvariable
Latency16.6s median
Run targetfal/ideogram/v4
Effective requestmodel: v4 · size: 1024x1024 · seed: 42
SourcesIdeogram model docsIdeogram pricing
Text57%Spatial66%Human61%Pro Studio65%Graphical49%Truth52%Preference34%Latency45%

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

57%
Typography Style67%
Writing accuracy42%

Spatial Reasoning

66%
Attributes Binding44%
Compositionality67%
Counting56%
Negation67%
Relative Position58%
Scale & Proportions44%

Human realism

61%
Faces & Expressions50%
Full Body33%
Hands33%
Multi-Subject67%

Professional Studio

65%
Camera & Lighting58%
Color Precision58%
Photorealism33%

Graphical design

49%
Data Visualisation67%
Layout & Design44%
Style Diversity42%

Truthfulness

52%
Photorealism67%
Physics & Reflections42%
World Knowledge42%