ImageBench

vs

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.

fal/microsoft/mai-image-2.5

API

MAI-Image-2.5 is Microsoft AI’s flagship photorealistic text-to-image and editing model, built for high-quality generation and controllable, identity-preserving edits at production scale. Strong at readable embedded text, composition, and studio/product visuals; ships a faster MAI-Image-2.5-Flash variant. Places top-3 on Arena.ai T2I. Note: a strict content filter refuses some benign prompts (e.g. involving children).

Maker
Microsoft
Family
MAI-Image
Model Size
not disclosed
unknown
Cost
~$0.05 / image
medium
Run Target
fal/microsoft/mai-image-2.5
Effective Request
aspect_ratio: 1:1 · num_images: 1 · output_format: png

fal/qwen-image-2-pro

API

Qwen Image 2.0 Pro is Alibaba's Qwen (Tongyi Lab) flagship hosted text-to-image model, a top-10 model on the Artificial Analysis image arena known for best-in-class typography, infographics, and poster/layout rendering. Distinct from the older local Qwen-Image checkpoint on this board; released ~April 2026.

Maker
Alibaba (Qwen)
Family
Qwen Image 2.0
Model Size
not disclosed
unknown
Cost
$0.075 / image
high
Run Target
fal/fal-ai/qwen-image-2/pro/text-to-image
Effective Request
image_size: square_hd · seed: 7
68.0vs66.9
Overall
69%vs71%
Capability
66.8vs63.1
Est. Preference
25.4svs12.8s
Avg Latency
Text Rendering80%100%Spatial Reasoning77%75%Human realism55%38%Truthfulness44%59%Professional Studio85%93%Graphical design79%88%Preference67%63%Latency5%25%

All 192 generations

Text Rendering80%vs100%

fal/microsoft/mai-image-2.5fal/qwen-image-2.0-pro

Typography Style100%vs100%

Writing accuracy75%vs100%

Spatial Reasoning77%vs77%

fal/microsoft/mai-image-2.5fal/qwen-image-2.0-pro

Attributes Binding89%vs89%

Compositionality89%vs89%

Counting56%vs67%

Negation78%vs89%

Relative Position83%vs83%

Scale & Proportions67%vs38%

Human realism55%vs38%

fal/microsoft/mai-image-2.5fal/qwen-image-2.0-pro

Faces & Expressions83%vs91%

Full Body0%vs18%

Hands75%vs0%

Multi-Subject67%vs50%

Truthfulness44%vs59%

fal/microsoft/mai-image-2.5fal/qwen-image-2.0-pro

Photorealism67%vs33%

Physics & Reflections42%vs50%

World Knowledge42%vs75%

Professional Studio85%vs93%

fal/microsoft/mai-image-2.5fal/qwen-image-2.0-pro

Camera & Lighting75%vs83%

Color Precision92%vs100%

Photorealism100%vs100%

Graphical design79%vs87%

fal/microsoft/mai-image-2.5fal/qwen-image-2.0-pro

Data Visualisation33%vs67%

Layout & Design89%vs78%

Style Diversity83%vs100%