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.

openai/gpt-image-2.5-sunburst

API

GPT Image 2.5 Sunburst is one of two API variants OpenAI released alongside ChatGPT Images 2.5 (September 2026), built for workflows where editing precision matters most.

Maker
OpenAI
Family
GPT Image
Model Size
not disclosed
unknown
Cost
token-based
high
Run Target
openai/gpt-image-2.5-sunburst
Effective Request
model: gpt-image-2.5-sunburst · size: 1024x1024 · quality: auto
82.1
Overall
88%
Capability
76.2
Est. Preference
169
Pass
23
Fail
16.9s
Avg Latency
9.3s
Min Latency
45.7s
Max Latency
Text Rendering100%Spatial Reasoning98%Human realism67%Truthfulness78%Professional Studio96%Graphical design96%Preference76%Latency17%

All 192 generations

Text Rendering100%

Typography Style100%

Writing accuracy100%

Spatial Reasoning98%

Attributes Binding100%

Compositionality100%

Counting100%

Negation100%

Relative Position100%

Scale & Proportions89%

Human realism67%

Faces & Expressions92%

Full Body33%

Hands67%

Multi-Subject83%

Truthfulness78%

Photorealism67%

Physics & Reflections75%

World Knowledge83%

Professional Studio96%

Camera & Lighting100%

Color Precision100%

Photorealism67%

Graphical design96%

Data Visualisation100%

Layout & Design89%

Style Diversity100%