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/moss-image-base

Local

MOSS-Image-Base is the OpenMOSS Team's Apache-2.0 1024 px text-to-image base model, with a Qwen3.5-9B text encoder and the FLUX.2 small-decoder VAE. It is the upstream checkpoint that Kromoss fine-tunes and that both 8-step turbo LoRAs were distilled from.

Maker: OpenMOSS Team
Family: MOSS-Image
Model Size: 5B
medium
Cost: local run; no API price
not_applicable
Run Target: gx10/moss-image-base
Effective Request: response_format: b64_json · size: 1024x1024 · num_inference_steps: 50 · guidance_scale: 3 · seed: 7
Sources: OpenMOSS-Team/MOSS-Image-Base model card MOSS-Image project page MOSS-Image technical report (arXiv 2610.23374)