Runs in the browser

MOSS-Image Generator

No ComfyUI and nothing to install. Type a brief, pick Base or SFT, and the first 3 images are free — no card. This studio runs the open MOSS-Image weights and is not affiliated with OpenMOSS.

3 free images, no card. Then $6 for 25, $18 for 100 or $45 for 300. One credit per image, 2K included, and credits never expire. Credit packs

MOSS-Image generator

Examples
Style preset
Aspect ratio
Resolution
2K output comes with any credit pack.
Advanced

Enter a whole number from 0 to 2147483647, or leave blank. The hosted API does not guarantee an identical image with the same seed.

Checking generator status…

New accounts get 3 free image credits

Your image will appear here

Write a prompt, quote any text you need, and press Generate.

Base or SFT: use SFT

One sentence covers it. SFT is the fine-tune of Base and the one to generate with; Base is the same model before that fine-tune, kept clean for LoRA training. Both cost the same here, and on short quoted text neither won our own test outright.

The full Base vs SFT comparison

Where it actually failed

These are from our own runs, not from the paper. Each one failed in a single, specific way.

A date read two ways

On a six-line jazz poster, one OCR pass read the date as "FRI 07 NOV" and another as "FRI D7 NOV". The zero in a date is the character to zoom in on.

The same prompt, a different word

The SFT run of that poster rendered "BLUE HOUR JAZZ" cleanly. Our first run of the identical prompt had the last word read as "JATT". Two runs can disagree, so generate two or three and proofread.

Long lines, not short ones

Short quoted lines, including mixed Chinese and English, came out clean across our tests. The errors sat in long addresses, small print and unusual punctuation.

The three failures, with the prompts

This generator vs ComfyUI

Both run the same open weights. The difference is who does the setup.

This generator vs ComfyUI
This pageComfyUI
SetupOpen the page and typeInstall ComfyUI, download the weights
Cost3 free images, then credit packsFree, on your own GPU
SettingsSteps and CFG are fixed for youYou set steps, CFG, shift and the sampler
ModelsBase and SFT in one selectorA loader change per checkpoint

More on the same model: The ComfyUI guide · ai flyer generator

Frequently asked questions about MOSS-Image

Do I need ComfyUI to use this MOSS-Image generator?

No. The generator above runs in your browser against the hosted MOSS-Image weights. ComfyUI is only needed if you want to run the open weights on your own machine.

Should the generator run Base or SFT?

SFT, unless you are comparing the two. It is the fine-tune and groups text more cleanly. Base is the checkpoint to train on, and it costs the same here.

How many images are free?

A new account gets 3 images, with no card required. After that, images come from one-time credit packs that never expire, and a failed generation returns its credit.

Will the text come out exactly as I typed it?

Short quoted lines usually do, including mixed Chinese and English. Long addresses, small print and unusual punctuation are where we have seen errors, so proofread anything you plan to publish.

Is this the official MOSS-Image generator?

No. It is an independent studio running the open weights, and it is not affiliated with the OpenMOSS team. The official repository is github.com/OpenMOSS/MOSS-Image.

Make something worth looking at

What do you want to say?

Start with a headline. Give it a little space. See where it takes you.

Open the studio