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.

fal/hunyuan-image-3

API

HunyuanImage 3.0 is Tencent's flagship open-weight text-to-image model — an 80B-parameter Mixture-of-Experts (~13B active), among the largest open image models. It emphasizes strong prompt following and legible in-image text, and reaches roughly ELO ~1125 on the Artificial Analysis image arena. Brings a new maker (Tencent) to the benchmark.

Maker: Tencent
Family: HunyuanImage 3.0
Model Size: 80B MoE (~13B active)
high
Cost: ~$0.09 / image
medium
Run Target: fal/fal-ai/hunyuan-image/v3/text-to-image
Effective Request: image_size: square_hd · seed: 7 · num_inference_steps: 28 · guidance_scale: 7.5
Sources: HunyuanImage 3.0 on fal HunyuanImage-3.0 (Tencent)