
Open-weights image model
MOSS-Image AI image generator
The open-source image model for long, detailed prompts and English or Chinese text, with community LoRAs.
No installation. No credit card. Private by default.
At a glance
MOSS-Image is an open-weights image model that Mage hosts on its own GPUs. On Mage it makes images from a prompt at up to 2K. It costs from 25 Gems (about $0.025) per image and runs unlimited from the Basic plan, and new accounts get 300 free Gems to start.
- 25Gems per image, from
- $0.025About, in US dollars
- BasicUnlimited from
- 2KTop resolution
Overview
What is MOSS-Image?
MOSS-Image is an open-source text-to-image model from the OpenMOSS team, released under the Apache 2.0 license. It follows long, detailed prompts closely and renders legible English and Chinese text.
Mage runs the full 50-step model on its own GPUs, so there is nothing to download and no GPU to rent, and community MOSS-Image LoRAs work on top of it.
Samples
Made with MOSS-Image
Generated on Mage with MOSS-Image. A prompt under a sample is the exact prompt that made it.

Chalkboard letteringPrompt: a cozy bookstore cafe at night, a chalkboard sign by the door reading "FRESH COFFEE $3", warm window light, rain on the street, photo

Chinese shop signPrompt: a carved wooden shop sign reading "老街茶馆" above a teahouse doorway in a misty old town, red lanterns glowing, stone steps wet from rain, evening, cinematic photo

A long, detailed scenePrompt: a lighthouse keeper in a heavy wool sweater standing at the top of a spiral staircase, holding a brass oil lamp, the great lens glowing behind him, storm clouds and crashing waves through the round windows, salt-stained glass, warm lamplight against cold blue dusk, 35mm film photograph, shallow depth of field
Strengths
What MOSS-Image does best
Best at
- Long, detailed prompts that describe a whole scene
- Signs, posters, and menus in English or Chinese
- Styles from community MOSS-Image LoRAs
- Clean photoreal portraits and illustration
Trade-offs
- No image input, Characters, or references
- Slower than few-step models such as Z-Image
Prompt guide
How to prompt MOSS-Image
Tips that work on this model. The prompts under the samples above are good places to start.
Tips
- Describe everything: subject, setting, light, camera, and style. MOSS-Image rewards detail.
- Put the exact words in quotes: a sign reading "FRESH COFFEE". Chinese text works the same way.
- Add a LoRA under Model for a consistent style. It starts at 0.5; raise it toward 1 if the style barely shows.
- Use the negative prompt for what keeps appearing that you do not want.
Specs
MOSS-Image specs and inputs
| Made by | Open weights, hosted on Mage's GPUs |
|---|---|
| Output | Images |
| Resolutions | 1K, 2K |
| Aspect ratios | 21:9, 16:9, 3:2, 5:4, 1:1, 4:5, 2:3, 9:16, 9:21 |
| Image inputs | None: it generates from the prompt alone |
| Video input | No |
| Mentions by @handle | No |
| Where to use it | The Mage app, the Mage assistant, Claude and ChatGPT through the Mage MCP server, and the Mage API |
Pricing
MOSS-Image price and plans
Every price comes from the same pricing that bills your generations.
| Model | Setting | Gems | Per | USD | Basic | Pro | Pro Plus | Max |
|---|---|---|---|---|---|---|---|---|
| MOSS-Image | 1K | 25 | image | $0.025 | Unlimited | Unlimited | Unlimited | Unlimited |
| MOSS-Image | 2K | 45 | image | $0.045 | Gems | Unlimited | Unlimited | Unlimited |
Prices in Gems, and in US dollars at the list price of the 10,000 Gems pack ($10). Each plan column shows what that plan runs unlimited; anything else costs Gems. Other settings, such as quality levels and extra input images, can change the price.
FAQ
MOSS-Image questions
How much does MOSS-Image cost?
Can I use MOSS-Image unlimited?
Can I try MOSS-Image for free?
What can I give MOSS-Image as input?
What resolutions does MOSS-Image support?
Can I use MOSS-Image through an API, or in Claude and ChatGPT?
Is what I make with MOSS-Image private?
Is MOSS-Image open source?
Can I use MOSS-Image LoRAs on Mage?
Can MOSS-Image write Chinese text?
Compare
More AI image models on Mage
Nano Banana 2.1
Google's image model for crisp text, many reference images, and edits written as one sentence.
SDXL Plus
Mage's high-detail SDXL pipeline, with faces and hands cleaned up automatically.
Stable Diffusion v1.5
The classic Stable Diffusion 1.5, on a budget and ready for older workflows.
Stable Diffusion v3.5 Large
The Stable Diffusion that follows detailed prompts most closely, at a budget price.
Stable Diffusion XL
Classic Stable Diffusion XL, tuned for photoreal output, at a budget price.
Z-Image
The fastest image model on Mage: photoreal drafts with solid anatomy and clean lettering.

Make your first image with MOSS-Image
New accounts get 300 free Gems, enough to try MOSS-Image. No install, no credit card.
Start creating for free