Plimo

Qwen Image 3.0: Make Posters, Infographics and Photo Edits with No Setup

Watercolor Pet
Neon City Poster
Hairstyle Try-on
Boxed Action Figure
Photo to Caricature
Cartoon Avatar
Ghibli Rainy Street
Evening Gown
Outfit Try-on
Studio Product Shot
Copper Lob Hairstyle
Studio Ghibli Filter
AI Palm Reading
Playful Caricature
Ghibli Wedding
Leather Jacket

What is Qwen Image?

Qwen Image is the image model line from Alibaba's Qwen team, known above all for putting real, readable text inside pictures, from a single headline to a page of small print. Plimo runs two versions of its third generation: 3.0 Pro for dense text and layouts, and the standard 3.0 for everyday pictures and quick edits.

  • A watercolor lighthouse on a cliff with the words Hello Sea painted on the rocks

    Qwen Image 3.0 Pro

    Handles dense text and layouts, such as posters and infographics.

    • 1K / 2K
    • Up to 3 reference photos

    From 15 credits

  • A watercolor lighthouse on a cliff with the words Hello Sea painted on the rocks

    Qwen Image 3.0

    Everyday pictures and quick edits from up to 3 photos.

    • 1K / 2K
    • Up to 3 reference photos

    From 10 credits

A short history of Qwen's image models

  1. August 2025

    Qwen Image

    A 20B open model that drew multi-line English and Chinese text most models of the time garbled.

  2. August 2025

    Qwen Image Edit

    The editing version could change what was in a photo, or rewrite its words in the original font.

  3. September 2025

    Edit 2509

    This update could edit from one to three photos at once, such as a person with a product.

  4. December 2025

    Qwen Image Edit 2511, Qwen Image 2512 and Layered

    These releases brought less drift in edits, more lifelike people, and a model that splits a picture into transparent layers.

  5. February 2026

    Version 2.0

    It merged generating and editing into one model, with prompts of up to 1,000 tokens.

  6. July 2026

    Version 3.0

    It raised the prompt limit to about 4,500 tokens and drew readable text down to ten pixels in twelve languages.

  7. September 2026

    Qwen Image 2.1

    A smaller 7B open model, it outputs transparent images directly and takes up to 10 reference images.

How to make an image with Qwen on Plimo

  1. Write the picture, words included

    The generator opens on 3.0 Pro. Describe the scene and put every word you want printed inside quotes, along with where it goes: title at the top, date in the corner.

  2. Add photos to edit instead

    Sign in and drop up to 3 photos next to the prompt, then say what should change. The shape you picked stays put even after photos are added.

  3. Choose a shape and size

    Seven shapes are on offer, from square to tall 9:16 and wide 16:9, plus 1K and 2K sizes in the settings beside the prompt.

  4. Generate and read it closely

    The Generate button shows the credits first. When the image lands in My creations, zoom in on the text; if a word slipped, press Remake and fix the prompt.

Why run Qwen on Plimo

  • No deployment, no weights to download

    Running Qwen models yourself usually means a big GPU, a ComfyUI graph and plenty of patience. Here you type a prompt and press Generate, from a laptop or a phone.

  • Built for text in the picture

    Posters, menus, slides and infographic grids are where this line has always been strongest, so it's the first model to try when the words matter as much as the image.

  • Room for a very long prompt

    The prompt box here takes up to 4,500 characters, enough to list every heading, label and line of copy on a busy page and say where each one sits.

  • A price you can predict

    The larger size costs the same as the smaller one on both versions. Each reference photo adds a little, and the Generate button always shows the total before you commit.

  • Members can use it for work

    Images made as a member come without a watermark and can go on a client deck, a shop window or a product page.

How Qwen compares with other image models

  • Nano Banana

    Google's Nano Banana is quick, keeps faces consistent across edits that mix up to 6 photos, and its version 2 works with free credits; Qwen's 3.0 Pro is aimed squarely at text-heavy layouts.

  • GPT Image

    GPT Image 2.5 from OpenAI is also good with lettering and adds transparent PNGs and up to 16 photos, and free credits run it at low quality; Qwen takes 3 photos and focuses on dense pages of text.

  • Grok Imagine

    xAI's Grok Imagine makes vivid pictures and also short videos with sound, which Qwen doesn't do; for edits, its image models take 3 or 5 photos.

  • Wan

    Wan, also from Alibaba, covers both images and video: Wan 2.7 Image Pro blends up to 9 photos, while Qwen stays with still pictures and puts its effort into text.

  • FLUX

    Black Forest Labs' FLUX 3 Image follows up to 10 reference photos and offers more shapes, including 21:9, so it suits detailed scenes more than pages of small print.

  • Seedream

    ByteDance's Seedream comes in three tiers that all follow up to 10 reference photos, and its Flash tier runs on free credits; pick Qwen when the layout is mostly words.

What people make with Qwen

  • A product box in three languages

    Write the product name, a short description and the ingredients in English, Spanish and Japanese, and ask for a clean box front. Alibaba says version 3.0 renders text in twelve languages, so each block stays in its own script.

  • A mock newspaper front page

    Give a masthead, a headline, two short stories and a photo caption, and ask for a front page in a classic broadsheet style. Long prompts are welcome, so spell out every column.

  • A study sheet with formulas

    List the formulas for a physics unit and a one-line note under each, then ask for a tidy revision sheet. Alibaba's own launch examples for 3.0 included multi-line math notation.

  • Fixing the words in a photo

    Upload a picture of a shop sign or a birthday banner and describe the new wording. Editing has been part of this line since 2025, and the current models carry that skill forward.

  • A comic strip with speech bubbles

    Describe four panels, who is in each and exactly what they say. Keep the lines short and the dialogue comes out in clean, readable bubbles.

  • A practice quiz on one page

    Describe a quiz with a header for name and date, ten numbered questions and an answer box under each. It saves a teacher or a parent from laying out the page by hand.

  • A quick mockup for a pitch

    Describe an app screen or a landing page with its buttons and headings to get a rough visual for a slide, before anyone opens a design tool.

  • A product on a new background

    Add up to 3 photos of a bottle or a bag, describe the setting, and ask for a short tagline in the corner, so the shot and the copy arrive together.

Questions about Qwen on Plimo

No. Plimo is an independent website that runs Alibaba's 3.0 and 3.0 Pro image models through an API, alongside models from other companies. We have no affiliation with Alibaba or the Qwen team.

These two models need a membership. New accounts get 100 free credits once, which for pictures work on Plimo Image 1.0, GPT Image 2.5, Nano Banana 2 and Seedream 5.0 Flash, so you can try the generator before picking a plan on the pricing page.

Look at the version cards higher up to see where each price starts; the button itself shows what this particular image will cost. The size you pick doesn't change it, while each added reference photo raises it a little.

They share the same settings and take the same photos. Pro is the one for crowded layouts like posters and infographics, while 3.0 costs less and suits everyday pictures and simple edits.

Yes. Add up to 3 photos in the generator and describe the change, like a new background, a different outfit or new words on a sign. There's no separate edit model to load.

No. Everything runs on Plimo, so there's no GPU, Python setup or ComfyUI workflow, and you don't need a key from Alibaba. Plimo doesn't offer its own API, so it's a place to make images, not to plug into your code.

Alibaba has published open weights for several earlier models in the line on Hugging Face and ModelScope. Version 3.0 launched without downloadable weights, and on Plimo you simply use it in the browser.

Alibaba says version 3.0 renders text in twelve languages, and English and Chinese have been the line's strengths from the start. Short phrases in quotes give the most reliable spelling.

Seven: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2 and 2:3. Your choice holds even after you add reference photos, so a square stays square.

Images you generate as a member have no watermark, and you're free to use them in paid projects. Non-members see a small mark in the corner, and their images are for personal use.

More image tools on Plimo

All models