How do I edit an image with Qwen Image 2?

Send the same request you would for text-to-image, but include image_url pointing at a public HTTPS image (no base64 or data: URIs — upload to your own storage first). That single source image switches Qwen Image 2 into editing mode, where your prompt describes the change: background swap, object edit, restyle, or text rewrite.

How good is Qwen Image 2 at rendering text in images?

In-image text is the headline strength of Qwen Image 2. It renders structured, legible type — readable headlines, labeled infographic steps, comic speech bubbles — where generic image models typically produce garbled letters. If your output has to carry words, that is the reason to choose Qwen Image 2.

How does Qwen Image 2 pricing work?

Credit-based: 1 credit equals $0.001 USD. Qwen Image 2 charges a flat rate per generated image, independent of resolution and aspect ratio, and each request produces one image. Failed and rejected requests are refunded automatically. See the pricing card above for the current per-image rate.

What aspect ratios and resolution does Qwen Image 2 support?

Text-to-image accepts 1:1, 3:4, 4:3, 9:16, and 16:9; editing adds 2:3, 3:2, and 21:9. The default is 16:9. Qwen Image 2 outputs up to 2048 × 2048 (2K), in PNG (default) or JPEG.

Does Qwen Image 2 support multiple images per request?

No — Qwen Image 2 returns exactly one image per request and editing takes a single source image. To produce a set, send multiple requests. Each one is billed as one image, so cost scales linearly with how many you submit.

Is Qwen Image 2 async?

Yes. You submit a Qwen Image 2 task and receive a task_id, then poll /api/v1/tasks/:id until it is completed. This is the same task flow as every other reApi model, and the API returns image URLs that you should mirror to your own storage if you need them long term.

Where is the full API reference for Qwen Image 2?

The complete parameter list, validation rules, error codes, and runnable cURL / Python / Node / Go examples for Qwen Image 2 live at /docs/qwen-image-2.

Is Qwen Image 2 free or paid?

Paid. Completed generations are charged at the rates in this page's pricing table. Eligible new accounts may receive promotional test credits; availability and amount are shown at signup or in the dashboard. Technical or provider failures are refunded automatically, except where this page documents a charged output-moderation block.

Does Qwen Image 2 require a monthly subscription?

No. Pricing is pay-as-you-go: $1 buys 1,000 credits and credits never expire. Cost scales with what you generate rather than with a recurring fee.

How do I get a Qwen Image 2 API key?

Sign up and create a key in the dashboard, no card required. Pass it as a Bearer token in the Authorization header.

Qwen Image 2 — Text-to-Image, Editing & In-Image Text

Qwen Image 2 is Alibaba's unified image model: write a prompt for text-to-image, or pass one source image to edit it in plain language. Its standout is structured in-image text — posters, slides, and infographics with type that actually reads — and you call Qwen Image 2 through reApi's unified API.

Highlights: Text rendering in zh & en · png / jpeg output
Best for: Posters, signage, bilingual text art
Input: Text, plus one reference image
Output: One image per request

Input

Prompt*

Up to 800 characters · required

Source image

playground-input.webp

Replace

✓ 1 file added

https://cdn.deepytb.com/landing/qwen-image-2/playground-input.webp

1/1 file

Public HTTPS URL only (no base64) · single image · leave empty for text-to-image

Aspect ratio

1:1, 3:4, 4:3, 9:16, 16:9 for text-to-image · editing also allows 2:3, 3:2, 21:9 · default 16:9

Output format

PNG (default) or JPEG

Seed

Same seed + prompt gives a repeatable result

Content filter

On in the playground; configurable via the API (API default: off).

Estimated cost~$0.03131 credits

Result

Try one of these prompts

Pricing

Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.

First test cost

$0.03131 credits

One sample at the cheapest tier (1 image).

Testing budget guide

Add credits

$10

≈ 324 tests

$50

≈ 1623 tests

$100

≈ 3246 tests

Per image: $0.031; Price · 1 image

What is the Qwen Image 2 API?

Qwen Image 2 is Alibaba's unified image model, exposed on reAPI as one async endpoint: send a prompt alone for text-to-image, or attach a single source image URL to edit it. Prompts run up to 800 characters, text-to-image accepts five aspect ratios while editing adds 2:3, 3:2 and 21:9 for eight in total, and each request returns one PNG or JPEG image, billed per image.

Qwen Image 2 sample generations and edits with in-image text

What you can build with this model

Real-world workflows and production use cases you can build and ship with this model.

Qwen Image 2 photorealistic portrait generated from a text prompt

Photoreal people, from a prompt

Qwen Image 2 is at its best with people. Describe someone and it returns a lifelike portrait — natural skin texture, believable expression, real lighting — instead of the waxy, uncanny faces most image models produce. Set the aspect ratio, write the look you want, and Qwen Image 2 hands back a photoreal person ready to use.

Open the playground

Qwen Image 2 before-and-after portrait edit of the same person

Edit people in plain language

Pass one photo of a person and a sentence, and Qwen Image 2 edits it while keeping the same face and identity — change the outfit, swap the background, adjust the expression, or restyle the whole shot. There is no mask to paint: attaching image_url is what tells Qwen Image 2 to edit instead of generate.

Qwen Image 2 consistent character across a nine-pose grid

One character, every pose

Lock a character once and Qwen Image 2 holds the same face, hair, and outfit across many poses and angles in a single request. It is built for pose sheets, lookbooks, and character references where consistency is the whole point — so the person in the ninth frame is unmistakably the person in the first.

Why reAPI for Qwen Image 2

Generate and edit in one model

Qwen Image 2 does both jobs through the same request to /api/v1/images/generations. Send a prompt for text-to-image; add a source image and the same call edits it. The mode is implicit in your inputs, so integrating Qwen Image 2 is one endpoint, not two — no branching client code, no second model to learn.

Text rendering that's actually legible

The reason teams reach for Qwen Image 2 is the in-image text. It renders structured, correctly-spelled type for posters, slides, infographics, and comics — the place generic image models fall apart. If your output has to carry words, it is the model that keeps them readable.

Async-first, unified

Submit a Qwen Image 2 task, get a task_id back, poll until it completes. 1 credit equals $0.001 USD and you pay a flat rate per generated image, so cost is predictable. The JSON matches the OpenAI image-generations contract, so adding Qwen Image 2 is a model-id change, not a new integration.

Qwen Image 2 vs Nano Banana

Both generate and edit images from one API. Qwen Image 2 leans into legible in-image text and single-source editing with flat per-image pricing. Here is how the two compare on publicly documented behavior.

Capability

Qwen Image 2 on reAPI

Nano Banana

Generation + editing

One model, one request shape — Qwen Image 2 switches to editing automatically when you attach a source image.

Generation and editing in one model, prompt-driven.

In-image text

A core strength — structured, legible typography for posters, slides, and infographics.

Renders text, with quality that varies by layout complexity.

Source images for editing

Single source image plus a natural-language instruction.

Accepts one or more reference images for blended edits.

Aspect ratios

Five ratios for text-to-image; eight when editing, including 21:9 ultrawide.

A fixed set of common aspect ratios.

Resolution

Up to 2K (2048 × 2048).

Up to 4K output.

Pricing model

Flat per generated image, independent of resolution and aspect ratio; pay-as-you-go credits, no subscription.

Per-image pricing that scales with output resolution.

Comparison reflects publicly documented behavior at the time of writing. Model behavior and pricing can change; check the pricing card above and the API docs for current values.

Integrate Qwen Image 2 in three steps

step 01
Create an API key
Sign up at reApi and grab an API key. Free signup credits cover plenty of Qwen Image 2 generations — enough to ship a first poster or edit before you top up.
Open
step 02
Submit a generation
POST to /api/v1/images/generations with model: qwen-image-2, a prompt, and optional image_url / image_size / output_format. Qwen Image 2 returns a task_id immediately — add image_url to edit instead of generate.
Open
step 03
Poll the result
GET /api/v1/tasks/:id until status is completed. The Qwen Image 2 response carries the image URL; mirror it to your own storage if you need it long term, since generated links expire.
Open

docs/api/qwen-image-2

API reference

Drop-in code and the full parameter table.

View full reference

curl -X POST https://reapi.ai/api/v1/images/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-image-2",
    "prompt": "a vintage travel poster that reads \"VISIT MARS\" in bold retro type",
    "image_size": "16:9"
  }'

Parameter validation failed

Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.

Authentication or balance issue

401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.

Content or material rejected

Reference material and outputs pass automated moderation. Rejected tasks fail with a clear error and a full refund.

Task failed or timed out

A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.

Frequently asked questions

Common questions about this model.

Qwen Image 2 is Alibaba's unified image model for both generation and editing. It does text-to-image from a prompt and natural-language editing from a single source image, and it is known for structured in-image text rendering. On reApi, Qwen Image 2 ships on the unified /api/v1/images/generations endpoint under the model id qwen-image-2.