Qwen-Image β hosted API access
This repository is not the model weights β it documents running the Qwen-Image family through a hosted API on apimodels.app, a multi-model gateway where one key and one OpenAI-compatible endpoint reach 170+ image, video, audio and language models.
- Original model & weights: https://huggingface.co/Qwen/Qwen-Image
- Authors: the Qwen team (Alibaba)
- License: Apache-2.0 (the Qwen-Image weights linked above)
Independent hosted-API integration. Not affiliated with or endorsed by the authors; does not redistribute their weights. All credit for the model is theirs.
Which version the API serves
apimodels.app does not run the original Qwen-Image checkpoint linked above. It serves Qwen Image 3.0, a newer version of the same family, as a hosted model in two tiers:
qwen3-imageβ standard tierqwen3-image-proβ higher-fidelity tier
We found no Qwen Image 3.0 weights in the Qwen organization on Hugging Face (checked November 2026), so the weights link above points to the open Qwen-Image release.
What it does
Text-to-image and image editing under one model name: send a prompt for a new image, or add 1 to 3 reference
images and the same model edits them. It is strong at dense, readable in-image text and structured layouts.
Output is 1K or 2K, in eight aspect ratios (1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16, 21:9) or auto.
A sample produced through the apimodels.app API.
Use it via API
Get a key at apimodels.app, then:
# text-to-image (async): create, then poll
curl -X POST https://api.apimodels.app/v1/images/generations \
-H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" \
-d '{ "model": "qwen3-image", "prompt": "a cozy bookstore cafe, warm light", "resolution": "1K", "aspect_ratio": "16:9" }'
# -> { "data": { "taskId": "...", "state": "pending" } }
curl "https://api.apimodels.app/v1/images/generations?task_id=TASK_ID" \
-H "Authorization: Bearer YOUR_API_KEY" # state -> completed, resultUrls[]
To edit instead, add "image_urls": ["https://example.com/photo.jpg"] (1 to 3 images, jpeg/png/webp, up to
10 MB each). Optional fields: negative_prompt, prompt_extend (default true), seed, callback_url.
Generation takes a while. On our production traffic the standard tier has a median of about 73 seconds; Pro
has a median of about 121 seconds with a long tail past 5 minutes. Allow about 3 minutes for standard and at
least 8 minutes for Pro before giving up on a poll, or pass callback_url and we POST the result to you.
Pricing
| Tier | 1K | 2K |
|---|---|---|
| qwen3-image | $0.035 / image | $0.035 / image |
| qwen3-image-pro | $0.037 / image | $0.075 / image |
| each reference image (editing, both tiers) | +$0.004 | +$0.004 |
Failed jobs are not charged. Result files are kept for 30 days; download anything you want to keep.
Links
- Run it: apimodels.app/models/qwen3-image
- API docs: apimodels.app/docs/qwen3-image
- Original model: https://huggingface.co/Qwen/Qwen-Image
More browser demos, a dataset and API guides: apimodels.app API demos collection
