Last updated: October 13, 2026

Qwen Image 2.1 Turbo API: pricing and code

Alibaba made qwen-image-2.1-turbo and qwen-image-2.1-pro available on Alibaba Cloud Model Studio alongside the open weights. Turbo is the cheaper, higher-throughput option. This page compares the official price with third-party hosts and shows how to call it.

Official price per image

List prices before promotions. Both models take text and image input and return images; batch inference is not supported.

Source: Model Studio: qwen-image-2.1-turbo

Source: Model Studio: qwen-image-2.1-pro

What 1,000 images cost

Source: Model Studio: qwen-image-2.1-turbo

Source: Model Studio: qwen-image-2.1-pro

Source: Runware model page

Source: Venice model page

Call it on Alibaba Cloud Model Studio

1. Create a Model Studio workspace in the region you want and an API key for it.
2. Save the key as DASHSCOPE_API_KEY.
3. POST to the multimodal-generation endpoint with model qwen-image-2.1-turbo. Size uses an asterisk, for example 2048*2048.
4. Download the returned image URL promptly: it expires after 24 hours.

# Singapore (international) region. Beijing: replace ap-southeast-1 with cn-beijing
# and use a Beijing API key. Keys and endpoints are not interchangeable.
curl -X POST "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" \
  -H "Authorization: Bearer $DASHSCOPE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-image-2.1-turbo",
    "input": {
      "messages": [
        { "role": "user", "content": [ { "text": "A red paper lantern on a rainy street at night, the sign reads \"Noodles\"" } ] }
      ]
    },
    "parameters": {
      "size": "2048*2048",
      "prompt_extend": true,
      "watermark": false
    }
  }'
# The image URL is at output.choices[0].message.content[0].image and expires after 24 hours.

Alibaba’s image API reference shows this request format with qwen-image-2.0 and 3.0 model names and does not yet list 2.1 Turbo by name; the model ID comes from the Turbo model page. prompt_extend (on by default) lets the service rewrite your prompt; set it to false for tighter control. We have not run this call ourselves.

Source: Model Studio image API reference

Source: Model Studio: qwen-image-2.1-turbo

Call it on Runware

Runware exposes the model as alibaba:[email protected]. Width and height can be any values from 192 to 4096 px (0.26–4.19 MP, 1:8 to 8:1), up to 10 reference images are accepted, and includeCost returns the charge with each result.

curl -X POST "https://api.runware.ai/v1" \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[{
    "taskType": "imageInference",
    "taskUUID": "6f1c2a1e-1d55-4a8e-9a39-0b6b8d6b2c11",
    "model": "alibaba:[email protected]",
    "positivePrompt": "A red paper lantern on a rainy street at night",
    "width": 1536,
    "height": 1024,
    "includeCost": true
  }]'

Source: Runware model page

Turbo or Pro?

Alibaba has not published a quality comparison between Turbo and Pro, so run the same 20–30 prompts through both before you decide.

• Pick Turbo. High-volume generation, previews, thumbnails, product variations and anything where 120 requests a minute matters.
• Pick Pro. Final hero images where you want the top hosted tier and can live with 20 requests a minute.
• Self-host. Research or personal use on your own GPU, where the open weights cost nothing per image.

Qwen Image 2.1 Turbo overview · Run it locally

Source: Model Studio: qwen-image-2.1-turbo

Source: Model Studio: qwen-image-2.1-pro

Independent guide, not affiliated with Alibaba or the Qwen team. Prices, licences and file names change; check the linked official pages before you build on them.