Skip to main content
Image models

Qwen Image 3 — Text to image API

Text to image with Qwen Image 3: request parameters, examples and response handling.

← Qwen Image 3

Endpoint: POST https://api.higgsfield.ai/alibaba/qwen-image-3/text-to-image

Endpoint ID: alibaba/qwen-image-3/text-to-image

▶Open API PlaygroundQwen Image 3 · Text to image↗

#Usage notes

  • prompt must contain non-whitespace text. The schema imposes no character maximum; its 4,500-token guidance is advisory.
  • enable_thinking defaults to true and requires prompt_extend=true. Set both to false to disable enhancement.
  • prompt_extend_mode supports direct or agent. No reference images are accepted by this endpoint.
  • resolution is a tier; 2k square maps to 1536 by 1536 pixels, not 2048 by 2048.

#Quick start

cURL
curl --request POST \
  --url https://api.higgsfield.ai/alibaba/qwen-image-3/text-to-image \
  --header "Authorization: Key ${HF_API_KEY_ID}:${HF_API_KEY_SECRET}" \
  --header "Content-Type: application/json" \
  --header "Idempotency-Key: ${IDEMPOTENCY_KEY}" \
  --data '{
  "prompt": "A ceramic vase on a wooden table beside a sunlit window, soft natural shadows, editorial photograph.",
  "resolution": "1k",
  "aspect_ratio": "1:1"
}'

#Input schema

seed integer

Optional random seed for similar, reproducible results. Minimum: 0. Maximum: 2147483647.

prompt string required

Positive prompt or editing instruction in Chinese or English. Alibaba recommends at most 4,500 tokens. Minimum characters: 1.

resolution string (default: 1k)

Output resolution tier. Both tiers produce PNG images. Allowed values: "1k", "2k".

aspect_ratio string (default: 1:1)

Output aspect ratio. Exact provider-recommended dimensions are selected automatically. Allowed values: "1:1", "2:3", "3:2", "3:4", "4:3", "7:9", "9:7", "9:16", "16:9", "21:9".

prompt_extend boolean (default: true)

Improve the prompt before generation. Required when thinking mode is enabled.

enable_thinking boolean (default: true)

Use model reasoning to improve image quality. Requires prompt enhancement.

negative_prompt string

Content, styles, or artifacts to avoid in the output.

prompt_extend_mode string (default: direct)

Direct enhancement is available for every request; agent enhancement is text-to-image only. Allowed values: "direct", "agent".

Complete JSON schema
json
{
  "type": "object",
  "allOf": [
    {
      "if": {
        "required": [
          "enable_thinking"
        ],
        "properties": {
          "enable_thinking": {
            "const": true
          }
        }
      },
      "then": {
        "required": [
          "prompt_extend"
        ],
        "properties": {
          "prompt_extend": {
            "const": true
          }
        }
      }
    }
  ],
  "title": "Qwen Image 3 Text to Image",
  "required": [
    "prompt"
  ],
  "properties": {
    "seed": {
      "type": "integer",
      "maximum": 2147483647,
      "minimum": 0,
      "description": "Optional random seed for similar, reproducible results."
    },
    "prompt": {
      "type": "string",
      "minLength": 1,
      "description": "Positive prompt or editing instruction in Chinese or English. Alibaba recommends at most 4,500 tokens."
    },
    "resolution": {
      "enum": [
        "1k",
        "2k"
      ],
      "type": "string",
      "default": "1k",
      "description": "Output resolution tier. Both tiers produce PNG images."
    },
    "aspect_ratio": {
      "enum": [
        "1:1",
        "2:3",
        "3:2",
        "3:4",
        "4:3",
        "7:9",
        "9:7",
        "9:16",
        "16:9",
        "21:9"
      ],
      "type": "string",
      "default": "1:1",
      "description": "Output aspect ratio. Exact provider-recommended dimensions are selected automatically."
    },
    "prompt_extend": {
      "type": "boolean",
      "default": true,
      "description": "Improve the prompt before generation. Required when thinking mode is enabled."
    },
    "enable_thinking": {
      "type": "boolean",
      "default": true,
      "description": "Use model reasoning to improve image quality. Requires prompt enhancement."
    },
    "negative_prompt": {
      "type": "string",
      "description": "Content, styles, or artifacts to avoid in the output."
    },
    "prompt_extend_mode": {
      "enum": [
        "direct",
        "agent"
      ],
      "type": "string",
      "default": "direct",
      "description": "Direct enhancement is available for every request; agent enhancement is text-to-image only."
    }
  },
  "additionalProperties": false
}

#Response

Submission returns a request handle. Poll status_url until status is completed, or use webhooks.

Accepted
{
  "status": "queued",
  "request_id": "REQUEST_ID",
  "status_url": "https://api.higgsfield.ai/requests/REQUEST_ID/status",
  "cancel_url": "https://api.higgsfield.ai/requests/REQUEST_ID/cancel"
}

A completed response includes the following output fields:

json
{
  "status": "completed",
  "request_id": "REQUEST_ID",
  "images": [
    {
      "url": "https://example.com/output.png"
    }
  ]
}

See idempotent requests, polling, request errors and authentication for shared request handling.

Was this page helpful?
Navigation