Qwen Image 3.0 API
qwen3-imageAlibaba Qwen Image 3.0 — text-to-image and image editing in one model, chosen automatically by whether the request carries reference images. Prompts up to 4.5k tokens with dense in-image text layout, 10px small-text rendering, 12 languages and 20+ native fonts. Output 1K or 2K in eight aspect ratios (1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16, 21:9), PNG. Flat $0.035 per image at either resolution; each reference image adds $0.004. Async — create then poll.
Generate Image
Result
Generated image will appear here
Enter a prompt and click Generate
One Model, Two Modes
Text-to-image and image editing on one model name — add reference images and it edits, omit them and it generates
Dense In-Image Text
Up to 4.5k-token prompts and 10px small-text rendering — newspapers, storyboards, menus and exam sheets in a single pass
12 Languages, 20+ Fonts
Native multilingual and multi-font rendering, legible for infographics and UI mockups straight out
Flat, Predictable Pricing
$0.035 per image at 1K or 2K; each reference image adds $0.004. Failures are never charged
Best For
High-volume posters, web and UI layouts, and text-heavy designs where small type must stay legible
About Qwen Image 3.0
Qwen Image 3.0 is an Image Generation API provided by Alibaba. Alibaba Qwen Image 3.0 — text-to-image and image editing in one model, chosen automatically by whether the request carries reference images. Prompts up to 4.5k tokens with dense in-image text layout, 10px small-text rendering, 12 languages and 20+ native fonts. Output 1K or 2K in eight aspect ratios (1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16, 21:9), PNG. Flat $0.035 per image at either resolution; each reference image adds $0.004. Async — create then poll. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: 1K: $0.035, 2K: $0.035.
Related model: qwen3-image-pro — the higher-fidelity tier of the same model — micro-expressions, pores and hair strands near real photography, 1K $0.037 / 2K $0.075
Related model: gpt-image-2 — stronger on general in-image text and the only one here with native 4K — 1K $0.025, 2K $0.03, 4K $0.05
Related model: nanobananapro — Gemini 3 Pro Image — reach for it when the subject is a realistic portrait rather than a text-heavy layout
Key Features
- One Model, Two Modes -- Text-to-image and image editing on one model name — add reference images and it edits, omit them and it generates
- Dense In-Image Text -- Up to 4.5k-token prompts and 10px small-text rendering — newspapers, storyboards, menus and exam sheets in a single pass
- 12 Languages, 20+ Fonts -- Native multilingual and multi-font rendering, legible for infographics and UI mockups straight out
- Flat, Predictable Pricing -- $0.035 per image at 1K or 2K; each reference image adds $0.004. Failures are never charged
- Best For -- High-volume posters, web and UI layouts, and text-heavy designs where small type must stay legible
Use Cases
Dense In-Image Text
Up to 4.5k-token prompts and 10px small-text rendering — newspapers, storyboards, menus and exam sheets generate as one structured layout, not a picture with text pasted on.
12 Languages, 20+ Fonts
Native multilingual and multi-font rendering, so infographics, UI mockups and multilingual marketing assets come out legible and usable straight away.
Structured Layouts
Understands complex instructions and organizes many elements into web pages, dashboards, game and livestream UIs in a single pass.
Edit From a Reference
Pass reference images in image_urls and it edits — swap elements, restyle, or repaint while keeping the subject; omit them for pure text-to-image. One model, both modes.
Predictable Pricing
Flat $0.035 per image at 1K and 2K, plus $0.004 per reference image. Failures are never charged, and one apimodels.app key also calls image, video, LLM and audio.
Why APIMODELS
- Unified API -- One API key to access all models, no need to register on multiple platforms
- Transparent Pricing -- Pay-as-you-go with per-call pricing published on every model page, ideal for indie developers and startups
- Instant Access -- Start using immediately after signup, supports Stripe and Alipay payments
- Full Documentation -- Detailed API docs with code examples in cURL, Python, and Node.js
Frequently Asked Questions
How much does Qwen Image 3.0 cost?
Qwen Image 3.0 is available through APIMODELS at: 1K: $0.035, 2K: $0.035. Billing is pay-as-you-go — you only pay for what you generate.
How to use Qwen Image 3.0 API?
Sign up at APIMODELS, get your API key, and call our unified API endpoint. We provide detailed API documentation with code examples in cURL, Python, and Node.js.
What is the difference between APIMODELS and the official Alibaba API?
APIMODELS offers the same Qwen Image 3.0 model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
What is Qwen Image 3.0?
Qwen Image 3.0 is Alibaba's image generation model, strong at realistic images, structured layouts, dense in-image text and high-quality typography. On apimodels.app text-to-image and image editing are one model, chosen automatically by whether you pass reference images — omit them and it generates, include them and it edits. 1K/2K, flat $0.035 per image, plus $0.004 per reference image.
How is Qwen Image 3.0 different from the Pro tier?
Standard Qwen Image 3.0 is the everyday tier: clear instruction following (prompts up to 4.5k tokens), stable text rendering (10px small text, 12 languages, 20+ fonts), flat $0.035 at 1K or 2K — built for posters, web pages and UI mockups at volume. Pro (qwen3-image-pro) is the higher-fidelity tier: micro-expressions, pores and hair strands near real photography, 1K $0.037 / 2K $0.075 — for hero shots, portraits and layouts that need the extra detail.
How is Qwen Image 3.0 priced?
Per image by resolution, pay-as-you-go: standard is $0.035 at both 1K and 2K; each reference image (for editing) adds $0.004. Failures are never charged. Example: a 1K text-to-image is $0.035; editing with one reference image is $0.035 + $0.004 = $0.039.
How does image editing work on Qwen Image 3.0?
Put reference images in image_urls and the backend routes to image-to-image automatically. Pass several to swap elements, change style, or repaint while keeping the subject. Omit image_urls for pure text-to-image. One model name, both modes — no endpoint switching.
How good is Qwen Image 3.0 at in-image text?
This is one of its standout strengths: up to 4.5k-token prompts and dense in-image layout let it generate newspapers, storyboards, menus and exam sheets in a single pass, with 10px small text rendered legibly. 12 languages and 20+ fonts render natively, so infographics and UI mockups are usable straight out — no post-hoc text overlays.
Which languages and fonts does it support?
Native rendering across 12 languages and 20+ fonts, with stable mixed-script and multilingual UI layouts. Prompts accept both Chinese and English. That makes it a good fit for multilingual marketing assets, educational content and web / game / livestream interface mockups.
What resolutions and aspect ratios are supported?
1K or 2K, in eight aspect ratios: 1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16, 21:9. Output is PNG. Reference images are up to 10MB each, in common formats (jpeg / png / webp).
How many reference images can I send?
Up to 3. Multiple references support multi-image fusion or tighter constraints when editing. Each reference image costs $0.004, so fusing three 1K images is $0.035 + 3×$0.004 = $0.047.
Is Qwen Image 3.0 synchronous or async?
Async: create a task, get a taskId, and poll — or pass a callBackUrl for a completion webhook. Results are transloaded to our own storage and returned as stable download links.
Qwen Image 3.0 vs gpt-image-2 or Nano Banana — which should I pick?
Qwen Image 3.0 leads on dense in-image text, multilingual typesetting and structured layouts (newspapers / menus / UI / exam sheets), at a lower price (from $0.035). gpt-image-2 is stronger on general in-image text and native 4K. Nano Banana (Gemini) is handy for realistic portraits and fast iteration. For text-dense, layout-heavy, multilingual work, reach for Qwen Image 3.0.
Where do I call the Qwen Image 3.0 API?
Sign up at apimodels.app, grab one API key, set model to qwen3-image — the same key also calls image, video, LLM and audio models. No juggling multiple providers or separate accounts; failures are never charged, billing is per image with no monthly fee.
How long does generation take, and is a long "pending" a stuck task?
Measured percentiles from our own production traffic: the standard tier has a median of 73s, with 90% finishing inside 96s and the fastest seen at 36s. A 3-minute polling window is plenty. If you cut the client request off after a few tens of seconds, the "stuck" task you see is your own timeout — the job is still running on our side and polling will flip state to completed. A genuine problem shows up as state=failed with a reason in failMsg.
Am I charged for failures, and can I just retry?
No charge. Only state=completed is billed; a failed task costs nothing and you can retry the exact same request. The most common failure is an upstream busy signal (UPSTREAM_BUSY), which usually clears on retry. Content-moderation and invalid-parameter failures will not improve on retry — change the prompt or the parameters instead.
How long do the result URLs stay alive?
Result files are kept for 30 days and then deleted automatically. Download and re-host them yourself if you need them longer — do not treat our URLs as permanent image hosting in your own database or pages.
Does it do 4K? Does it support mask-based inpainting?
Neither, to be blunt. No Qwen Image 3.0 tier goes above 2K — for native 4K use gpt-image-2 ($0.05 per 4K image). There is also no mask/inpainting parameter: the way to edit a region is to pass the original through image_urls and describe the change in the prompt. It follows that kind of instruction well, but it is not pixel-level mask control.
What goes in the model field — is this the same as Alibaba's qwen-image-3.0?
Same model, different name. Our model string is qwen3-image (or qwen3-image-pro), which maps to Alibaba's qwen-image-3.0 and qwen-image-3.0-pro. Send ours, with no prefix.
Can the 10px small text go straight to production?
Legible is not the same as character-accurate. The small text it renders really is sharp, but spelling, formulas, prices, barcodes and dates — anything where one wrong character matters — need a human pass before you ship. Use it as a typesetting capability, not as a text-entry tool.
How does Qwen Image 3.0 compare to other Image Generation models?
On APIMODELS, Qwen Image 3.0 runs alongside 60+ models on one API key and one balance, so choosing is about fit, not lock-in. It supports Text to Image, Image Editing, Dense Text Layout, 12 Languages, 20+ Fonts, 1K/2K, Async, and you can weigh it on price and capability against other Image Generation models, then switch by changing a single model-name string — no new account or integration. Browse every Image Generation option with live pricing at apimodels.app/models.
What can Qwen Image 3.0 do?
Qwen Image 3.0 supports: Text to Image, Image Editing, Dense Text Layout, 12 Languages, 20+ Fonts, 1K/2K, Async. See the APIMODELS docs for full parameters and call examples.
Can I access the Qwen Image 3.0 API from anywhere (incl. China)?
Yes. APIMODELS exposes Qwen Image 3.0 through a single unified API and one key — no separate provider accounts, and no need to handle each provider's regional network access yourself.
What payment methods are supported?
We support Stripe (Visa, Mastercard, and other international cards) and Alipay. Credits are available instantly after payment.
Related models
GPT Image 2.5 prompt examples
See all 238 →Prompts shared by their authors — copy and adapt them. Each one credits its author and links back to the original post.
Weathered sailor on a fishing boat
Create a photorealistic candid photograph of an elderly sailor standing on a small fishing boat. He has weathered skin with visible wrinkles…
by OpenAI
Automatic coffee machine workflow infographic
Create a detailed Infographic of the functioning and flow of an automatic coffee machine like a Jura. From bean basket, to grinding, to scal…
by OpenAI
Thread streetwear ad with exact typography
Give me a cool in culture ad / fashion shot for a brand called Thread. It's a hip young street brand. The ad shows a group of friends hangin…
by OpenAI
Need prompts?
We curate copy-ready prompt libraries — every entry shows its full text and a sample result, ready to adapt.
Related reading
How to get access, regional availability, and how this model compares with its alternatives.
Comparisons
How this model stacks up against the alternatives on results, price and features.