Qwen Image 2.1: AI Image Generator Online
Bottom line
Qwen Image 2.1 generates images online with a real transparent PNG/WebP background option: 3 credits at 1K, 5 at 2K, 9 aspect ratios, 5,000-character prompts. Pick Qwen Image 2.1 in the Text to Image tool, write your prompt, and pay per image in credits. The price is visible before you generate.
Qwen Image 2.1 is Alibaba's Qwen 2.1 text-to-image model, and you can use it online on VdoBloom with no API key: type a prompt, pick 1K or 2K and one of nine aspect ratios, and generate. It costs 3 credits at 1K and 5 credits at 2K, so the 10 free signup credits cover three 1K images.
Its headline feature is a real transparent background. Set the background to transparent and the output format to PNG or WebP, and the image comes back with an alpha channel, ready to drop onto a website, a slide or a product mock-up without a separate background-removal step. It also accepts prompts of up to 5,000 characters, has an enhance-prompt toggle (on by default) that expands short prompts, and is the Qwen generator with the tall 9:21 ratio.
This is the page for generating with Qwen Image 2.1. Its photo-editing sibling, with mask inpainting, is Qwen Image 2.1 Edit; the Qwen 3 generators are compared in the table below.
Qwen Image 2.1 specs at a glance
Values below come from VdoBloom's Qwen Image 2.1 service (which follows KIE's published rules for the model), the generator configuration and the credit table.
| Spec | Qwen Image 2.1 on VdoBloom |
|---|---|
| What it does | Generates images from a text prompt (text-to-image) |
| Price | 3 credits at 1K, 5 credits at 2K (credit keys qwen2-1-t2i-1k and qwen2-1-t2i-2k) |
| Resolutions | 1K (default) and 2K; no 4K |
| Aspect ratios | 9: 3:2 (default), 1:1, 4:3, 3:4, 2:3, 16:9, 9:16, 21:9, 9:21 |
| Prompt length | Up to 5,000 characters |
| Background | Opaque or transparent (real alpha channel) |
| Output format | PNG (default), WebP or JPEG; transparent needs PNG or WebP |
| Enhance prompt | On by default; rewrites a short prompt into a fuller scene. Turn it off for literal control |
| Negative prompt | Not available on this model |
| Provider | KIE (qwen2-1/text-to-image) |
| Content label | HOT (all Qwen models) |
| On VdoBloom since | 22 September 2026 |
How to get a transparent PNG from Qwen Image 2.1
Transparency needs two settings together. JPEG has no alpha channel, so VdoBloom rejects a transparent background with JPEG output instead of quietly returning an opaque image. With PNG or WebP selected, the model renders the subject on a genuinely empty background.
Describe only the subject. A prompt that asks for a scene ('a cup on a wooden table in a cafe') gives the model something to paint behind the cup; a prompt that names the object, its material and its lighting ('a ceramic coffee cup, matte sage glaze, soft studio light from the left, centered') leaves the background empty. Keep enhance-prompt in mind too: it expands short prompts into fuller scenes, so turn it off if it keeps adding a setting you did not ask for.
- Open the image generator and select Qwen Image 2.1.
- Set Background to transparent.
- Set Output format to PNG (or WebP for smaller web files).
- Write a subject-only prompt; consider turning enhance-prompt off.
- Generate at 1K (3 credits) to check the cut-out, then at 2K (5 credits) for the final.
What Qwen Image 2.1 costs, worked out
Resolution sets the price; aspect ratio, background and format do not. Dollar figures use the USD Lite pack: 300 credits for $19.50 one-time (never expire) or $15 a month (subscription credits reset each cycle).
| Job | Credits | One-time Lite rate | Monthly Lite rate |
|---|---|---|---|
| 1 image at 1K | 3 | about $0.20 | $0.15 |
| 1 image at 2K | 5 | about $0.33 | $0.25 |
| 4 transparent icon drafts at 1K + 1 at 2K | 17 | about $1.11 | $0.85 |
| A set of 12 stickers at 2K | 60 | $3.90 | $3.00 |
| 1K images per 300-credit Lite pack | 100 | $19.50 | $15 |
| Free signup credits (10) | 3 images at 1K (1 credit left) | $0 | $0 |
Qwen Image 2.1 against the other Qwen models
VdoBloom runs six current Qwen image models, and Qwen Image 2.1 is the generator row with a transparent background, the longest prompt allowance and the 9:21 ratio. It is also the cheapest Qwen generator at 1K. At 2K it costs the same 5 credits as Qwen Image 3.0, so the choice there is features: 2.1 for transparency, long prompts and 9:21; 3.0 for the negative-prompt field and the Qwen 3 model. Each sibling has its own page, linked below.
If you start from a photo instead of a prompt, the matching editor is Qwen Image 2.1 Edit, which adds mask inpainting and 1 to 10 reference images at the same 3 or 5 credits.
| Model | Job | 1K | 2K | Prompt cap | Aspect ratios | Inputs | What only it does |
|---|---|---|---|---|---|---|---|
| Qwen Image 3.0 | Text to image | 5 credits | 5 credits | 800 characters | 8 | Prompt + optional negative prompt | Same price at 2K as at 1K |
| Qwen Image 3.0 Pro | Text to image | 6 credits | 7 credits | 800 characters | 8 | Prompt + optional negative prompt | Pro realism and typography tier |
| Qwen Image 3.0 Edit | Edit photos | 5 credits | 5 credits | 800 characters | 8 | Up to 10 uploads (first 3 used on the primary provider) | Flat-price edits at 2K |
| Qwen Image 3.0 Pro Edit | Edit photos | 6 credits | 7 credits | 800 characters | 8 | Up to 10 uploads (first 3 used on the primary provider) | Pro-tier edits for lettering and fine detail |
| Qwen Image 2.1 | Text to image | 3 credits | 5 credits | 5,000 characters | 9 (adds 9:21) | Prompt + enhance-prompt toggle | Real transparent PNG/WebP background |
| Qwen Image 2.1 Edit | Edit photos | 3 credits | 5 credits | 5,000 characters | 9 + auto | 1 to 10 references, or 1 photo + painted mask | Mask inpainting and transparent output |
Read more: Qwen Image 3.0 · Qwen Image 3.0 Pro · Qwen Image 3.0 Edit · Qwen Image 3.0 Pro Edit · Qwen Image 2.1 · Qwen Image 2.1 Edit
Where it runs on VdoBloom
Qwen Image 2.1 is in the image generator (Images, then Generate): pick the Qwen family in the model picker, then Qwen Image 2.1. The composer shows the aspect-ratio menu, the 1K/2K switch, the output format, the background toggle and the enhance-prompt toggle, and the cost (3 or 5 credits) shows before you generate. Results save to My Creations. The Qwen Image 2.1 tool page that used to exist was folded into this model page, so this is the place to start.
- Image generator: /ai-image-generator/
- Model picker path: Qwen family, then Qwen Image 2.1
- Defaults: 1K, 3:2, PNG, opaque background, enhance-prompt on
- Price shown before you generate; failed jobs are refunded
Prompting notes: long prompts and enhance-prompt
With 5,000 characters available, Qwen Image 2.1 can take a full brief: subject, materials, lighting direction, camera height, lens feel, palette and the exact words of any sign or label. Write it in plain sentences, one idea per sentence, and say where things sit in the frame for the ratio you chose.
Enhance-prompt is on by default and works in the other direction: it expands a short prompt into a fuller scene before generation. That helps when you type five words and want a finished picture. It gets in the way when you want exactly what you wrote, or a clean transparent cut-out, so switch it off in those cases.
- Short idea, want a finished scene: leave enhance-prompt on.
- Detailed brief or exact layout: turn enhance-prompt off.
- Transparent assets: subject-only prompt, PNG or WebP, enhance-prompt usually off.
- Text in the image: put the exact words in quotes and keep them short.
- Use 9:21 for tall story graphics and 21:9 for website banners.
Limits and weak cases
What Qwen Image 2.1 does not do, from the code, and which page covers it.
- Transparent background only with PNG or WebP; JPEG plus transparent is rejected.
- No negative-prompt field; Qwen Image 3.0 and 3.0 Pro have one.
- No 4K; the top tier is 2K.
- No reference images on this model; to edit a photo or use a mask, switch to Qwen Image 2.1 Edit.
- Enhance-prompt can add scenery you did not ask for; turn it off for literal prompts.
What creators use Qwen Image 2.1 for
- Logos, stickers, icons and product cut-outs as transparent PNG or WebP, with no background to remove afterward.
- Cheap 1K drafts at 3 credits before a 5-credit 2K final.
- Long, detailed scene prompts of up to 5,000 characters, or a short prompt expanded by enhance-prompt.
- Tall 9:21 and cinema-wide 21:9 frames for story graphics and website hero strips.
How to use Qwen Image 2.1 on VdoBloom
- 1Open the Text to Image tool and select Qwen Image 2.1.
- 2Write your prompt.
- 3Set resolution (1K/2K), aspect ratio, review the credit cost, and generate.
Frequently asked questions
Can I use Qwen Image 2.1 online?
Yes. On VdoBloom it runs in the browser in the image generator: select Qwen Image 2.1, write a prompt and generate. There is no API key and nothing to install, and new accounts get 10 free credits.
How much does Qwen Image 2.1 cost?
3 credits at 1K and 5 credits at 2K. Aspect ratio, background and format do not change the price. On the one-time Lite pack (300 credits for $19.50) a 1K image is about 20 cents.
Can I try Qwen Image 2.1 for free?
Yes. The 10 free signup credits cover three images at 1K or two at 2K.
Does Qwen Image 2.1 make transparent PNGs?
Yes. Set Background to transparent and Output format to PNG or WebP, and the image has a real alpha channel. JPEG cannot hold transparency, so that combination is rejected.
Why does my transparent image still have a background scene?
Usually the prompt describes a setting, or enhance-prompt expanded a short prompt into a scene. Describe only the subject and turn enhance-prompt off.
Which aspect ratios does Qwen Image 2.1 support?
Nine: 3:2 (default), 1:1, 4:3, 3:4, 2:3, 16:9, 9:16, 21:9 and 9:21, at 1K or 2K.
What is the prompt limit for Qwen Image 2.1?
5,000 characters, against 800 on the Qwen 3 models.
What does the enhance-prompt toggle do?
It rewrites your prompt into a fuller scene description before generation. It is on by default; turn it off when you want the model to follow exactly what you typed.
Qwen Image 2.1 vs Qwen Image 3.0: which should I use?
For transparent backgrounds, long prompts, the 9:21 ratio or the cheapest 1K image (3 credits), use 2.1. For a negative-prompt field on the Qwen 3 model, use 3.0. At 2K both cost 5 credits.
Can Qwen Image 2.1 edit photos?
This model generates from text. Its sibling Qwen Image 2.1 Edit takes 1 to 10 reference images, or one photo plus a painted mask for local edits, at the same 3 or 5 credits.
Qwen Image 2.1 guides, prices and comparisons
This page is the Qwen Image 2.1 hub. Each guide below goes deeper on one question and links back here.
AI transparent background images with Qwen Image 2.1
Step-by-step transparent PNG and WebP generation.
Qwen Image 2.1 guide
Mask inpainting with the Edit model, plus the generator's settings.
AI background remover
For photos you already have, where generating a new cut-out is not an option.
Multi-model AI image generator
Compare one prompt across Qwen and other model families.