fal/google/nano-banana-2 Benchmark — 73/100 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.

MakerGoogle

FamilyNano Banana

Model sizenot disclosedundisclosed

Cost$0.08 / imagemid

Latency28.1s median

Run targetfal/google/nano-banana-2

Effective requestmodel: nano-banana-2 · size: 1024x1024 · seed: 42

SourcesGoogle model docsGoogle 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

83%

Typography Style67%

Writing accuracy75%

Spatial Reasoning

65%

Attributes Binding44%

Compositionality33%

Counting67%

Negation44%

Relative Position67%

Scale & Proportions67%

Human realism

65%

Faces & Expressions50%

Full Body58%

Hands67%

Multi-Subject33%

Professional Studio

79%

Camera & Lighting67%

Color Precision75%

Photorealism67%

Graphical design

66%

Data Visualisation67%

Layout & Design78%

Style Diversity50%

Truthfulness

84%

Photorealism100%

Physics & Reflections67%

World Knowledge100%