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/flux-2-klein-9b

Local

The local FLUX.2 [klein] 9B run uses Black Forest Labs' larger small-model checkpoint, prioritizing better quality while staying in the low-latency FLUX.2 Klein family.

Maker: Black Forest Labs
Family: FLUX.2 [klein]
Model Size: 9B
high
Cost: local run; no API price
not_applicable
Run Target: gx10/flux2-klein-9b
Effective Request: Effective request fields unknown
Sources: Black Forest Labs FLUX.2 overview FLUX.2-klein-9B model card

56.6

Overall

53%

Capability

60.1

Est. Preference

102

Pass

90

Fail

8.5s

Avg Latency

8.1s

Min Latency

9.2s

Max Latency

All 192 generations

Text Rendering47%

Typography Style100%

Writing accuracy33%

Spatial Reasoning60%

Attributes Binding67%

Compositionality89%

Counting33%

Negation44%

Relative Position67%

Scale & Proportions56%

Human realism24%

Faces & Expressions50%

Full Body17%

Hands8%

Multi-Subject17%

Truthfulness41%

Photorealism33%

Physics & Reflections50%

World Knowledge33%

Professional Studio89%

Camera & Lighting92%

Color Precision92%

Photorealism67%

Graphical design67%

Data Visualisation0%

Layout & Design56%

Style Diversity92%