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/krea-2-turbo

Local

Krea 2 Turbo is Krea AI's 8-step distilled open-weight Krea 2 checkpoint, built for fast, high-quality text-to-image inference while pairing with RAW-trained LoRAs.

Maker: Krea AI
Family: Krea 2 Open-Source
Model Size: 12B
high
Cost: local run; no API price
not_applicable
Run Target: gx10/krea2-turbo
Effective Request: response_format: b64_json · size: 1024x1024 · num_inference_steps: 8 · guidance_scale: 0 · seed: 7
Sources: Krea 2 Open-Source page Krea 2 Turbo model card

56.5

Overall

58%

Capability

55.2

Est. Preference

111

Pass

81

Fail

73.2s

Avg Latency

70.9s

Min Latency

262.9s

Max Latency

All 192 generations

Text Rendering60%

Typography Style100%

Writing accuracy50%

Spatial Reasoning74%

Attributes Binding89%

Compositionality89%

Counting89%

Negation44%

Relative Position83%

Scale & Proportions44%

Human realism21%

Faces & Expressions42%

Full Body8%

Hands8%

Multi-Subject33%

Truthfulness44%

Photorealism33%

Physics & Reflections50%

World Knowledge42%

Professional Studio85%

Camera & Lighting83%

Color Precision83%

Photorealism100%

Graphical design67%

Data Visualisation0%

Layout & Design56%

Style Diversity92%