ImageBench

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 ImageBench V1 run.

local/z-image-6b

Local

Z-Image 6B is part of Tongyi-MAI's efficient Z-Image model family, a 6B text-to-image line designed for practical generation quality, instruction following, and efficient deployment.

Maker
Tongyi-MAI / Alibaba
Family
Z-Image
Model Size
6B
high
Cost
local run; no API price
not_applicable
Run Target
gx10/z-image-6b
Effective Request
response_format: b64_json · seed: 7
Sources
Tongyi-MAI Z-Image repositoryZ-Image-Turbo model card
47.4
Overall
53%
Capability
41.7
Est. Preference
102
Pass
90
Fail
130.7s
Avg Latency
123.3s
Min Latency
146.4s
Max Latency
Text Rendering67%Spatial Reasoning63%Human realism26%Truthfulness26%Professional Studio85%Graphical design63%Preference42%Latency0%

All 192 generations

Text Rendering67%

Typography Style100%

Writing accuracy58%

Spatial Reasoning63%

Attributes Binding89%

Compositionality67%

Counting44%

Negation56%

Relative Position67%

Scale & Proportions56%

Human realism26%

Faces & Expressions67%

Full Body0%

Hands8%

Multi-Subject33%

Truthfulness26%

Photorealism0%

Physics & Reflections42%

World Knowledge17%

Professional Studio85%

Camera & Lighting83%

Color Precision83%

Photorealism100%

Graphical design63%

Data Visualisation0%

Layout & Design44%

Style Diversity92%