MOSS-Image AI Image Generator

Write English or Chinese into the image, place text and objects with a JSON layout, and follow a long prompt in illustrated or photographic styles at up to 2K.

Up to 2K
English and Chinese text
Text to image
OpenMOSS · MOSS-Image docs ↗

Made with MOSS-Image

Vintage travel poster titled Lisbon with the line Sun, Sea & Tram 28

Lisbon travel poster

Prompt

A vintage travel poster of a seaside town at sunset, bold hand-lettered title "Lisbon" at the top and the line "Sun, Sea & Tram 28" at the bottom, flat colours, textured paper, balanced layout.

Key Features of MOSS-Image

Bilingual text, a JSON layout, long prompts, and illustrated or photographic styles. Prefill any row; nothing runs until you generate.

English and Chinese text

Put both scripts in one frame. This fridge has a yellow Chinese grocery note and a blue English lesson note. MixVio returns a PNG at 1K or 2K.

Refrigerator with a yellow Chinese grocery sticky note and a blue English sticky note about English lessons

How to Use MOSS-Image

1
1. Start from a text brief
Open MOSS-Image with a text brief, or prefill a Key Feature or Example. Nothing runs until you generate.
Open MOSS-Image with a text brief, or prefill a Key Feature or Example. Nothing runs until you generate.
2
2. Write the scene, or paste a JSON layout
For placed text and objects, paste a JSON layout with an items array. For a scene, write English or Chinese in the prompt. This route does not take a reference image.
For placed text and objects, paste a JSON layout with an items array. For a scene, write English or Chinese in the prompt. This route does not take a reference image.
3
3. Choose ratio and resolution, then generate
Choose a ratio and 1K or 2K, set 1 to 4 images, confirm the credit estimate, then generate. Review the text before publishing. MixVio does not expose a quality menu, 4K, a seed, or a negative prompt.
Choose a ratio and 1K or 2K, set 1 to 4 images, confirm the credit estimate, then generate. Review the text before publishing. MixVio does not expose a quality menu, 4K, a seed, or a negative prompt.

Create with MOSS-Image ↑

MOSS-Image Pricing

A 1K square image is 4 credits and a 2K square image is 14 credits. The range is 2–14 credits per image. A 1K 16:9 image is the floor. Asking for more than one image multiplies the quote. Confirm the reservation before you generate.

Compare MOSS-ImageHow MOSS-Image compares to other models

Pick by the job: bilingual text and a JSON layout, photo edits, or native 4K with references. Specs below are the live MixVio routes.

MOSS-Image

This model

English and Chinese text, JSON layouts, and long prompts at 1K or 2K

1K–2K PNG2–14 credits per image

Qwen Image 3 →Alibaba

Dense layouts, readable multilingual text, and up to three photo refs

1K–2K PNG3–6 credits per output

GPT Image 2.5 →OpenAI

Keep-and-change edits, reference fidelity, and Flare or Sunburst

1K–4K PNG1–8 credits per output

Attribute	This modelMOSS-Image	AlibabaQwen Image 3 →	OpenAIGPT Image 2.5 →	Black Forest LabsFLUX 3 Image →
Best for	English and Chinese text, JSON layouts, and long prompts at 1K or 2K	Dense layouts, readable multilingual text, and up to three photo refs	Keep-and-change edits, reference fidelity, and Flare or Sunburst	Native 2K and 4K photos, readable text, and up to 10 photo refs
Text rendering	Excellent	Excellent	Strong	Strong
Image editing	No	Strong	Excellent	Strong
Multi-image references	No	Medium	Strong	Excellent
Photorealism	Strong	Strong	Excellent	Excellent
Speed	Medium	Medium	Excellent	Medium
Max resolution	1K–2K	1K–2K	1K–4K	1K–4K
Credits	2–14 credits per image	3–6 credits per output	1–8 credits per output	3–43 credits per output

Stay on MOSS-Image for English and Chinese text, a JSON layout, and long prompts at 1K or 2K, with up to four images and no reference photo. Stay on Qwen Image 3 to edit up to three photos. Stay on GPT Image 2.5 for keep-and-change edits, Flare or Sunburst, and up to 4K. Stay on FLUX 3 Image for native 2K and 4K with up to 10 photo references. Speed is the MixVio catalog tier, not a MixVio latency benchmark.

MOSS-ImageMOSS-Image guidefal MOSS-ImageQwen Image 3GPT Image 2.5FLUX 3 Image

MOSS-Image FAQs

What is MOSS-Image?

MOSS-Image is OpenMOSS’s text-to-image model. On MixVio you write a prompt, optionally as a JSON layout, and generate a PNG at 1K or 2K. It is for in-image English and Chinese text, placed layouts, and long prompts. It does not edit a photo or take a reference image.

What is MOSS-Image best for?

Use it when the words in the image matter: bilingual notes, posters, exam pages, and JSON-placed objects. The gallery includes illustrated and photographic styles. It does not replace an edit model when you need to change an existing photo.

Can I upload a reference or edit an image?

No. This route is text to image only. Switching to it clears attached images. Use Qwen Image 3, GPT Image 2.5, or FLUX 3 Image when the job starts from a photo.

Which sizes and how many images can I generate?

Choose 1K or 2K and 1:1, 4:3, 3:4, 3:2, 2:3, 16:9, or 9:16. One request can return 1 to 4 PNGs. 4K, 21:9, and the 512-pixel square preset are not available. Desktop opens at 1K and 1:1. Phone opens at 1K and 9:16.

How many credits does MOSS-Image use?

From 2 credits for a 1K 16:9 image to 14 credits for a 2K square image. The default 1K square image is 4 credits. The count multiplies when you ask for more than one image. Confirm the reservation before you generate. Failed runs cost no credits.

Does MOSS-Image rewrite my prompt?

A JSON prompt with an items array is sent as written, so placed boxes are not rewritten. Natural language and other JSON keep prompt expansion on. MixVio does not expose a seed, a quality menu, or a negative-prompt field. Sampling stays at 50 steps and guidance 3.

How does MOSS-Image compare with Qwen Image 3, GPT Image 2.5, and FLUX 3 Image?

Stay on MOSS-Image for bilingual in-image text and a JSON layout at 1K or 2K, with up to four images and no reference photo. Stay on Qwen Image 3 to edit up to three photos. Stay on GPT Image 2.5 for keep-and-change edits, Flare or Sunburst, and up to 4K. Stay on FLUX 3 Image for native 2K and 4K with up to 10 photo references. Specs in the table are the live MixVio routes.

Is MOSS-Image available on the Free plan?

Yes, when your balance covers the quote. Free-plan outputs are for evaluation and are not for commercial publishing. Paid-plan outputs may be used commercially, subject to the Terms of Service and current OpenMOSS terms. You must have rights to every prompt, face, and brand.

Are my prompts and outputs private on MixVio?

Yes. Prompts are routed server-side, and finished images are archived privately to your workspace. Keep private prompts and asset IDs out of shared links.

Create images with MOSS-Image.
Private. Credits shown first.

Write English or Chinese, paste a JSON layout, choose 1K or 2K, then generate with MOSS-Image selected.