# Qwen Image

> Alibaba's Qwen image model

## Quick Reference

- Model ID: qwen-image
- Creator: Alibaba
- Status: active
- Family: qwen
- Base URL: https://api.lumenfall.ai/openai/v1

## Specifications

- Input Modalities: text
- Output Modalities: image
- Supported Modes: Text to Image, Image Edit

## API Parameters

The compiled parameter schema for this model is available via the API: `GET /v1/models/qwen-image?schema=true`.

### Core Parameters

- `prompt` (string) — REQUIRED: Text prompt for image generation. Modes: Text to Image, Image Edit
- `negative_prompt` (string): Negative prompt to guide generation away from undesired content. Modes: Text to Image, Image Edit
- `seed` (integer): Random seed for reproducibility. Modes: Text to Image, Image Edit

### Size & Layout

- `size` (string): Image dimensions as WxH pixels (e.g. "1024x1024") or aspect ratio (e.g. "16:9"). Values: 1365x768, 768x1365, 1254x836, 836x1254, 887x1182, 1024x1024, 1183x887. Modes: Text to Image, Image Edit
- `aspect_ratio` (string): Aspect ratio of the output image (e.g. "16:9", "1:1"). Values: 9:16, 2:3, 3:4, 1:1, 4:3, 3:2, 16:9. Modes: Text to Image, Image Edit
- `resolution` (string): Output resolution tier (e.g. "1K", "4K"). Values: 1K. Modes: Text to Image, Image Edit

### Media Inputs

- `image` (file) — REQUIRED: Input image(s) to edit. Modes: Image Edit

### Output & Format

- `response_format` (string): How to return the image. Default: url. Values: url, b64_json. Modes: Text to Image, Image Edit
- `output_format` (string): Output image format. Values: png, jpeg, gif, webp, avif. Modes: Text to Image, Image Edit
- `output_compression` (integer): Compression level for lossy formats (JPEG, WebP, AVIF). Modes: Text to Image, Image Edit
- `n` (integer): Number of images to generate. Default: 1. Modes: Text to Image, Image Edit

### Additional Parameters

- `cfg_scale` (number): Classifier-free guidance scale — higher values stick more closely to the prompt. Modes: Text to Image, Image Edit
- `prompt_enhancement` (string): Whether an LLM rewrites/expands the prompt before generation (off, on). Values: off, on. Modes: Text to Image, Image Edit
- `strength` (number): How much to transform the input image: 0 keeps it unchanged, 1 fully regenerates from the prompt. Modes: Image Edit
- `acceleration` (string): Acceleration level for image generation. Options: 'none', 'regular', 'high'. Higher acceleration increases speed. 'regular' balances speed and quality. 'high' is recommended for images without text.. Values: high, none, regular. Modes: Text to Image. Only available via fal
- `disable_safety_checker` (boolean): Disable safety checker for generated images.. Modes: Text to Image, Image Edit. Only available via replicate
- `enable_safety_checker` (boolean): If set to true, the safety checker will be enabled.. Modes: Text to Image. Only available via fal
- `extra_lora_scale` (array): Scales for additional LoRAs as an array of numbers (e.g., 0.5, 0.7). Must match the number of weights in extra_lora_weights.. Modes: Text to Image, Image Edit. Only available via replicate
- `extra_lora_weights` (array): Additional LoRA weights as an array of URLs. Same formats supported as lora_weights (e.g., ['https://huggingface.co/flymy-ai/qwen-image-lora/resolve/main/pytorch_lora_weights.safetensors', 'https://huggingface.co/flymy-ai/qwen-image-realism-lora/resolve/main/flymy_realism.safetensors']). Modes: Text to Image, Image Edit. Only available via replicate
- `go_fast` (boolean): Run faster predictions with additional optimizations.. Modes: Text to Image, Image Edit. Only available via replicate
- `image_size` (string): Image size for the generated image. Values: optimize_for_quality, optimize_for_speed. Modes: Text to Image, Image Edit. Only available via replicate
- `lora_scale` (number): Determines how strongly the main LoRA should be applied.. Modes: Text to Image, Image Edit. Only available via replicate
- `lora_weights` (string): Load LoRA weights. Only works with text to image pipeline. Supports arbitrary .safetensors URLs, tar files, and zip files from the Internet (for example, 'https://huggingface.co/flymy-ai/qwen-image-lora/resolve/main/pytorch_lora_weights.safetensors', 'https://example.com/lora_weights.tar.gz', or 'https://example.com/lora_weights.zip'). Modes: Text to Image, Image Edit. Only available via replicate
- `loras` (array): The LoRAs to use for the image generation. You can use up to 3 LoRAs and they will be merged together to generate the final image.. Modes: Text to Image. Only available via fal
- `num_inference_steps` (integer): Number of denoising steps. Recommended range is 28-50, and lower number of steps produce lower quality outputs, faster.. Modes: Text to Image, Image Edit
- `output_quality` (integer): Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is lowest quality. Not relevant for .png outputs. Modes: Text to Image, Image Edit. Only available via replicate
- `replicate_weights` (string): Load LoRA weights from Replicate training. Only works with text to image pipeline. Supports arbitrary .safetensors URLs, tar files, and zip files from the Internet.. Modes: Text to Image, Image Edit. Only available via replicate
- `sync_mode` (boolean): If `True`, the media will be returned as a data URI and the output data won't be available in the request history.. Modes: Text to Image. Only available via fal
- `use_turbo` (boolean): Enable turbo mode for faster generation with high quality. When enabled, uses optimized settings (10 steps, CFG=1.2).. Modes: Text to Image. Only available via fal
- `watermark` (boolean): Whether to add the provider watermark to the generated media.. Modes: Text to Image. Only available via alibaba


## Model Identifiers

- Primary Slug: qwen-image

## Dates

- Released: August 2025

## Tags

image-generation

## Available Providers

### Replicate

- Config Key: replicate/qwen-image
- Provider Model ID: qwen/qwen-image
- Pricing: $0.025/image
  - Source: https://replicate.com/qwen/qwen-image

### fal.ai

- Config Key: fal/qwen-image
- Provider Model ID: fal-ai/qwen-image
- Pricing: $0.020/megapixel
  - Source: https://fal.ai/models/fal-ai/qwen-image

### Alibaba Cloud

- Config Key: alibaba/qwen-image
- Provider Model ID: qwen-image-plus
- Pricing: $0.030/image
  - Source: https://modelstudio.console.alibabacloud.com/ap-southeast-1?tab=doc#/doc/?type=model&url=2840914_2&modelId=qwen-image-plus


## Arena Benchmarks

### Chalkboard Menu

- Elo: 1152
- Record: 3W / 2L / 0T (5 battles)
- Rank: #11 of 69

### The Halloween Invitation

- Elo: 1147
- Record: 4W / 2L / 0T (6 battles)
- Rank: #9 of 70

### Candid Street Photography

- Elo: 1117
- Record: 3W / 1L / 0T (4 battles)
- Rank: #30 of 68

### Magic Burger Explosion: Fiery Photorealism Challenge

- Elo: 1070
- Record: 1W / 3L / 0T (4 battles)
- Rank: #48 of 69

### Apollo 11: Journey to Tranquility

- Elo: 1063
- Record: 1W / 3L / 0T (4 battles)
- Rank: #42 of 70


## Use Cases & Category Performance

### Art (Text-to-Image)

- Rank: #11 of 31
- Elo: 1184
- Record: 5W / 2L / 0T (7 battles)
- Win Rate: 71.4%

### Product, Branding & Commercial (Text-to-Image)

- Rank: #35 of 61
- Elo: 1188
- Record: 10W / 10L / 0T (20 battles)
- Win Rate: 50.0%

### Text Rendering (Text-to-Image)

- Rank: #40 of 61
- Elo: 1182
- Record: 10W / 11L / 0T (21 battles)
- Win Rate: 47.6%

### Photorealism (Text-to-Image)

- Rank: #41 of 62
- Elo: 1162
- Record: 10W / 12L / 0T (22 battles)
- Win Rate: 45.5%

### Creativity (Text-to-Image)

- Rank: #42 of 62
- Elo: 1173
- Record: 7W / 9L / 0T (16 battles)
- Win Rate: 43.8%

### Aesthetics (Text-to-Image)

- Rank: #43 of 62
- Elo: 1194
- Record: 12W / 13L / 0T (25 battles)
- Win Rate: 48.0%

### Prompt Adherence (Text-to-Image)

- Rank: #53 of 61
- Elo: 1168
- Record: 8W / 12L / 0T (20 battles)
- Win Rate: 40.0%


## Image Gallery

9 images available for this model. Browse all at https://lumenfall.ai/models/alibaba/qwen-image/gallery

### Curated Examples

- [A wide, cinematic shot of a meticulously detailed, handcrafted leather-bound journal lying on a r...](https://assets.lumenfall.ai/U6jPphhKw_bxl8dW8mVrCVekMugu1gTy1f7fxxb7HH4/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/q3esp9djd7p31j9czudq0qkhrcia@jpeg)
- [A close-up, cinematic macro shot of an weathered leather craftsman's workbench. In sharp focus ar...](https://assets.lumenfall.ai/hoc9kJDXVEskksxBxGodikFuR30V3lh9v3h5RuiQD5U/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/fwyl12wcr70bzg3e7jvsg7ysy65r@jpeg)
- [A hyper-realistic close-up of an elderly sculptor's hands working on a delicate clay bust. Fine d...](https://assets.lumenfall.ai/2GIzCedaxzwgJBSckVrs-tDYzRScilOdtNAKJFnbkr8/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/851ivvxbzh9ljkfxgg685swxw200@jpeg)
- [A sun-drenched Mediterranean balcony overlooking the sea, overflowing with vibrant bougainvillea ...](https://assets.lumenfall.ai/0RC6PC-B3nxLv8oTsMXo3euVq4oaR_8WVm0VhWt_4Gc/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/98fcqefagktq6km7bp14x0ijgx62@jpeg)

### Arena Competition Results

- [Chalkboard Menu](https://assets.lumenfall.ai/_9DG7KV6B87guhlK9lfpN4tp7G6kXyBw4ZkiqF-2lJ4/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/9n6sbaxo313vugk36ui0vj06t87w@jpeg): #11 of 69 (Elo 1152)
- [The Halloween Invitation](https://assets.lumenfall.ai/Ep6vv10jGKMNhLY50ukyI2jWDd2VYvi1AQdgAp6nI7A/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/8udf5kk5ih84kqcbtgdsvosmvust@jpeg): #9 of 70 (Elo 1147)
- [Candid Street Photography](https://assets.lumenfall.ai/pwOZ1CsTrzecaGadyri1u3VjVQr9QhavfpIaoWjkIaY/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/mc9oyslrtibhi95lfofwuriudc2p@jpeg): #30 of 68 (Elo 1117)
- [Magic Burger Explosion: Fiery Photorealism Challenge](https://assets.lumenfall.ai/u2hA181jAqiGeFzoP_VejWYw5qkmkz8I7xXM5FUx7I0/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/ar3s4f6xu5ao0q1lrz4y81v1vzox@jpeg): #48 of 69 (Elo 1070)
- [Apollo 11: Journey to Tranquility](https://assets.lumenfall.ai/KIjpwtGG3Bi6-6tChsOb__dqg8Ozeh0gav7a8MiuWg0/rs:fit:1500:1500/plain/gs://lumenfall-prod-assets/xq6alo77e1omzvr78oco6mpjy2uh@jpeg): #42 of 70 (Elo 1063)

## Example Prompt

The following prompt was used to generate an example image in our playground:

A sun-drenched Mediterranean balcony overlooking the sea, overflowing with vibrant bougainvillea and terracotta pots. In the soft background shadows near a wooden bench, a capybara naps peacefully while a breakfast spread sits on the foreground table.

## Code Examples

### Text to Image (/v1/images/generations)

#### cURL

curl -X POST \
  https://api.lumenfall.ai/openai/v1/images/generations \
  -H "Authorization: Bearer $LUMENFALL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-image",
    "prompt": "",
    "size": "1024x1024"
  }'

# Response:
# { "created": 1234567890, "data": [{ "url": "https://...", "revised_prompt": "..." }] }

#### JavaScript

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: 'YOUR_API_KEY',
  baseURL: 'https://api.lumenfall.ai/openai/v1'
});

const response = await client.images.generate({
  model: 'qwen-image',
  prompt: '',
  size: '1024x1024'
});

// { created: 1234567890, data: [{ url: "https://...", revised_prompt: "..." }] }
console.log(response.data[0].url);

#### Python

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://api.lumenfall.ai/openai/v1"
)

response = client.images.generate(
    model="qwen-image",
    prompt="",
    size="1024x1024"
)

# { created: 1234567890, data: [{ url: "https://...", revised_prompt: "..." }] }
print(response.data[0].url)

### Image Edit (/v1/images/edits)

#### cURL

curl -X POST \
  https://api.lumenfall.ai/openai/v1/images/edits \
  -H "Authorization: Bearer $LUMENFALL_API_KEY" \
  -F "model=qwen-image" \
  -F "image=@source.png" \
  -F "prompt=Add a starry night sky to this image" \
  -F "size=1024x1024"

# Response:
# { "created": 1234567890, "data": [{ "url": "https://...", "revised_prompt": "..." }] }

#### JavaScript

import OpenAI from 'openai';
import fs from 'fs';

const client = new OpenAI({
  apiKey: 'YOUR_API_KEY',
  baseURL: 'https://api.lumenfall.ai/openai/v1'
});

const response = await client.images.edit({
  model: 'qwen-image',
  image: fs.createReadStream('source.png'),
  prompt: 'Add a starry night sky to this image',
  size: '1024x1024'
});

// { created: 1234567890, data: [{ url: "https://...", revised_prompt: "..." }] }
console.log(response.data[0].url);

#### Python

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://api.lumenfall.ai/openai/v1"
)

response = client.images.edit(
    model="qwen-image",
    image=open("source.png", "rb"),
    prompt="Add a starry night sky to this image",
    size="1024x1024"
)

# { created: 1234567890, data: [{ url: "https://...", revised_prompt: "..." }] }
print(response.data[0].url)


## About

## Overview
Qwen Image is a text-to-image generation model developed by Alibaba Cloud’s Qwen team. It serves as the visual synthesis component of the broader Qwen ecosystem, designed to transform natural language prompts into high-fidelity imagery. The model is distinguished by its strong alignment with complex linguistic instructions and its ability to handle both English and Chinese prompts with high semantic accuracy.

## Strengths
*   **Multilingual Prompt Comprehension:** The model demonstrates superior performance in processing Chinese-language prompts, accurately capturing cultural nuances and idioms that Western-centric models often misinterpret.
*   **Compositional Accuracy:** It excels at spatial reasoning and multi-object placement, ensuring that elements described in a prompt maintain the correct relationship to one another.
*   **Text Rendering:** Qwen Image shows higher-than-average stability when generating legible text within images, such as signage, labels, or posters, reducing the common "gibberish" artifacts found in earlier diffusion models.
*   **Fine-Grained Detail:** The model is optimized for high-resolution output with a focus on realistic textures, particularly in skin tones, fabric weaves, and architectural materials.

## Limitations
*   **Anatomical Consistency:** Like many diffusion-based models, it can occasionally struggle with complex human anatomy, such as the specific number of digits on hands or complex overlapping limbs in action shots.
*   **Stylistic Range:** While versatile, the model tends toward a "digital photography" or "clean 3D render" aesthetic by default; achieving hyper-abstract or specific traditional art styles may require more intensive prompt engineering compared to models like Midjourney.

## Technical Background
Qwen Image belongs to the Qwen family of models, leveraging a large-scale diffusion transformer architecture tailored for high-dimensional visual synthesis. The training process involves a multi-stage pipeline that utilizes high-quality captioned image datasets, with a specific focus on cross-modal alignment between the Qwen LLM's text embeddings and the visual latent space. This allows the model to inherit the deep semantic understanding found in Alibaba's flagship language models.

## Best For
Qwen Image is particularly effective for marketing localization projects involving Chinese text, technical illustrations requiring precise object placement, and general-purpose asset generation for web and mobile interfaces. Its price point of $0.02 makes it a cost-effective choice for developers building high-volume image generation workflows.

Qwen Image is available for immediate deployment and testing through **Lumenfall’s unified API and playground**, allowing you to integrate its generative capabilities into your applications with minimal setup.

## Frequently Asked Questions

### How much does Qwen Image cost?

Qwen Image starts at $0.02 per image through Lumenfall. Pricing varies by provider. Lumenfall does not add any markup to provider pricing.

### How do I use Qwen Image via API?

You can use Qwen Image through Lumenfall's OpenAI-compatible API. Send requests to the unified endpoint with model ID "qwen-image". Code examples are available in Python, JavaScript, and cURL.

### Which providers offer Qwen Image?

Qwen Image is available through Replicate, fal.ai, and Alibaba Cloud on Lumenfall. Lumenfall automatically routes requests to the best available provider.

## Links

- Model Page: https://lumenfall.ai/models/alibaba/qwen-image
- About: https://lumenfall.ai/models/alibaba/qwen-image/about
- Providers, Pricing & Performance: https://lumenfall.ai/models/alibaba/qwen-image/providers
- API Reference: https://lumenfall.ai/models/alibaba/qwen-image/api
- Benchmarks: https://lumenfall.ai/models/alibaba/qwen-image/benchmarks
- Use Cases: https://lumenfall.ai/models/alibaba/qwen-image/use-cases
- Gallery: https://lumenfall.ai/models/alibaba/qwen-image/gallery
- Playground: https://lumenfall.ai/models/alibaba/qwen-image/playground
- API Documentation: https://docs.lumenfall.ai