qwen-image-3

Quick example

curl https://reapi.ai/api/v1/images/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-image-3.0-pro",
    "prompt": "A cafe menu board, three columns - appetizers, mains, desserts - five dishes each with prices, serif headings, cream paper background",
    "size": "3:4",
    "resolution": "2K",
    "n": 1
  }'

import requests

resp = requests.post(
    "https://reapi.ai/api/v1/images/generations",
    headers={
        "Authorization": "Bearer YOUR_API_KEY",
        "Content-Type": "application/json",
    },
    json={
        "model": "qwen-image-3.0-pro",
        "prompt": "A cafe menu board, three columns - appetizers, mains, desserts - five dishes each with prices, serif headings, cream paper background",
        "size": "3:4",
        "resolution": "2K",
        "n": 1,
    },
    timeout=30,
)
print(resp.json())

const r = await fetch("https://reapi.ai/api/v1/images/generations", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "qwen-image-3.0-pro",
    prompt: "A cafe menu board, three columns - appetizers, mains, desserts - five dishes each with prices, serif headings, cream paper background",
    size: "3:4",
    resolution: "2K",
    n: 1,
  }),
});
console.log(await r.json());

package main

import (
    "bytes"
    "encoding/json"
    "fmt"
    "io"
    "net/http"
)

func main() {
    body, _ := json.Marshal(map[string]any{
        "model":      "qwen-image-3.0-pro",
        "prompt":     "A cafe menu board, three columns - appetizers, mains, desserts - five dishes each with prices, serif headings, cream paper background",
        "size":       "3:4",
        "resolution": "2K",
        "n":          1,
    })
    req, _ := http.NewRequest("POST",
        "https://reapi.ai/api/v1/images/generations", bytes.NewReader(body))
    req.Header.Set("Authorization", "Bearer YOUR_API_KEY")
    req.Header.Set("Content-Type", "application/json")

    resp, _ := http.DefaultClient.Do(req)
    defer resp.Body.Close()
    out, _ := io.ReadAll(resp.Body)
    fmt.Println(string(out))
}

Submit response

{
  "id": "task_018f5a3a1b6e7d9f8c2b4d6e8f0a2c4e",
  "model": "qwen-image-3.0-pro",
  "status": "processing",
  "created_at": 1735000000
}

Poll GET /api/v1/tasks/{id} (see the Tasks reference) until status === "completed". The completed payload's output.image_urls holds the generated image URLs.

Authentication

Every call needs a Bearer token. Generate keys at reapi.ai/settings/apikeys.

Authorization: Bearer YOUR_API_KEY

Endpoint

POST /api/v1/images/generations
GET  /api/v1/tasks/{id}

Submission is async. The POST returns immediately with a task_id; the task endpoint returns the same envelope until completion. Polling does not consume credits.

Tiers

Two model ids share one parameter shape:

Pick Pro for dense text and layout work — menus, newspapers, storyboards, exam papers. Standard is the cheaper general-purpose tier and is the better default when the output is a single illustration or photo.

Modes

The mode is implicit in the request shape — there is no mode field:

Reference images are not billed, so an edit costs the same as a generation at the same tier and resolution.

Request body

model — string, required

qwen-image-3.0 or qwen-image-3.0-pro. See Tiers.

prompt — string, required

What to draw or how to edit. The upstream model accepts roughly 4.5k tokens, which is enough to describe a full multi-column page in one request. The limit is counted in tokens rather than characters, so an over-long prompt is rejected upstream rather than by a character count on reAPI's side.

image_urls — array, optional

1-3 public image URLs to edit or draw from. Supplying them switches the request to image-to-image.

Accepted source formats: JPG / JPEG / PNG / BMP / TIFF / WEBP / GIF, up to 10 MB each.

n — integer, default 1

How many images to return, 1-6. Every delivered image is billed, so n: 6 costs six images.

size — string, default 1:1

Aspect ratio, or an explicit pixel size. Documented ratios:

1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3

The field also accepts 16x9 style and explicit pixels such as 1600x900. For custom pixels each edge must be 512-2048 and the aspect ratio must fall between 1:8 and 8:1.

Ratio and tier combine as follows:

resolution — string, default 1K

1K or 2K (lowercase is accepted and normalised). On qwen-image-3.0 both tiers cost the same; on qwen-image-3.0-pro 2K costs twice 1K.

negative_prompt — string, optional

Content to keep out of the image.

prompt_extend — boolean, default false

Enables intelligent prompt rewriting before generation. Note this defaults to false here — the 2.0 series defaulted to true. Turn it on for terse prompts; leave it off when you have written a precise layout brief you want followed literally.

prompt_extend_mode — string, optional

How aggressively the prompt is rewritten. Requires prompt_extend: true.

Sending prompt_extend_mode without prompt_extend: true, or agent together with image_urls, returns a 400.

Pricing

Qwen Image 3.0 bills per successfully generated image. Total charge is the per-image rate for the selected tier and resolution, times n:

credits = ceil(per_image_usd × n × 1000)

where 1 credit = $0.001 USD. The rate depends on the model id and, on Pro, the resolution tier:

Reference images are free — an edit costs the same as a generation. Failed and rejected requests are not charged.

The exact per-image credit cost for each row surfaces on the model page and through the playground estimator before submit.

Response

The poll envelope returns image URLs in output.image_urls:

{
  "id": "task_019dfd44b7fd74168541552a3260a623",
  "model": "qwen-image-3.0-pro",
  "status": "completed",
  "output": {
    "image_urls": [
      "https://cdn.reapi.ai/...png"
    ]
  }
}

For n > 1, image_urls holds every delivered image. Mirror the URLs to your own storage if you need long-term retention.

Errors

Failures return the standard reAPI envelope { error: { code, message, request_id } }. Common cases:

• Invalid input (n outside 1-6, more than 3 image_urls, a custom pixel edge outside 512-2048, an invalid ratio, prompt_extend_mode without prompt_extend: true, agent mode with image_urls, a non-HTTPS media URL) → 400.
• Insufficient credits → 402.
• Rate limited → 429.

See the full catalog at /docs/api/errors.

Tips

• The mode is decided by your inputs, not a flag — omit image_urls for pure generation, include it to edit.
• Spend the prompt budget. This model is built for long, structural briefs: describe the masthead, each column, the captions and the sidebar in one request instead of generating pieces and assembling them.
• Draft at 1K and re-run only the winner at 2K. On the Pro tier that halves the cost of iteration.
• Leave prompt_extend off when your prompt is already a precise layout spec — rewriting can restructure a brief you wrote deliberately.
• Reach for qwen-image-3.0-pro when the image has to carry a lot of small text; use qwen-image-3.0 for general illustration and photography.

Related

• Qwen Image 2
• Wan 2.7 Image
• Tasks API
• Error codes