Qwen Image 2.1 LoRA API

Introduction

Qwen Image 2.1 LoRA Text to Image runs Alibaba's newest open-weight image model with your own LoRA weights loaded on top — up to three files per request, each with its own strength.

A LoRA here is a URL, not an upload.

Choose a Qwen Image 2.1 LoRA mode by starting material

Key features

No upload, no allow-list, no approval queue

Most hosted APIs that accept custom weights want them uploaded, reviewed, or picked from a curated list — which excludes exactly the weights most people want to run. Here you supply a direct link to a .safetensors file or a Hugging Face repository and a strength, and it is fetched when the request runs.

Three slots, independent strengths

Three entries per call, each with its own scale from 0 to 4, covers the stack production work actually uses: a character LoRA holding the person, a house-style LoRA at a lower weight colouring the frame, a detail LoRA for skin or fabric. Change one scale per run — moving two at once leaves you unable to attribute the result.

Request parameters

Pricing

Prices are in credits, the same credits you use on the website (1 credit = $0.01). The hold is taken when the task is accepted and settled to the actual charge when it finishes; failed tasks are refunded.

Each task is rounded up to whole credits, so one run at the lowest tier costs 5 credits per image ($0.05).

For an exact number before you run, send the same body to POST /jobs/quote.

Use cases

Consistent characters at scale

A character LoRA holds a face and body across hundreds of frames far more reliably than seeds or reference chaining.

House style over a general model

A style LoRA trained on your own catalogue makes output look like yours rather than like a generic render.

Production stacks

Character, style and detail weights together in one request, each at its own strength.

Prompt guide & pro tips

Let the weights do their job

Describe the scene, not the thing the LoRA already knows. Restating a character's features when a character LoRA is loaded tends to fight it rather than reinforce it.

Start at 1, then move one value

Scale 1 is neutral. Drop to 0.4-0.6 when a weight overwhelms the prompt; push past 1 when it has to survive a busy scene. One change per run.

Budget for the first call

The weights are fetched before generation. A 1.2 GB file added around 25 seconds to that request in our testing. Later calls with the same file are faster.

Check a new LoRA actually does something

Weights for a different base architecture can load without an error and barely affect the image. Run one obvious test prompt before committing a batch.

API quick start

Call the Qwen Image 2.1 LoRA API in three steps

1. Create an API key. Sign in to PopNow and create a key in the API console. Keys start with sk-popnow- and use the same credits as the website.
2. Send the request. POST /jobs/createTask with model alibaba/qwen-image-2.1-lora/text-to-image and an input object that follows the parameter table. The response returns a task ID straight away.
3. Get the result. Poll GET /jobs/recordInfo with the task ID, or pass a callBackUrl and we post the result to it. Download the output links when the task succeeds.

# 1. Create the task. Keep the Idempotency-Key if you retry after a network error.
export IDEMPOTENCY_KEY="$(uuidgen)"
curl -X POST https://api.popnow.ai/api/v1/jobs/createTask \
  -H "Authorization: Bearer $POPNOW_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $IDEMPOTENCY_KEY" \
  -d '{
  "model": "alibaba/qwen-image-2.1-lora/text-to-image",
  "input": {
    "prompt": "A ceramic teapot on a linen cloth, soft window light, muted palette",
    "aspect_ratio": "1:1",
    "resolution": "1k",
    "output_format": "jpeg"
  }
}'

# 2. Copy data.taskId from the response, then poll until state is succeeded, failed, canceled or expired.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.popnow.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
  -H "Authorization: Bearer $POPNOW_API_KEY"

Replace the prompt and any links with your own. The API tab has the response, polling and callback reference.

Qwen Image 2.1 LoRA API questions

All Qwen Image models

Plain Qwen Image 2.1 is cheaper when you don't need your own weights.

• Qwen Image 2.1 Pro APIUp to six detailed 2K images from one promptFrom 6 credits per image
• Qwen Image 2.1 Turbo APIFast 8-step renders, up to six per promptFrom 3 credits per image
• Qwen Image 2.1 APIOpen model with 2K output and transparent backgroundsFrom 4 credits per image
• Qwen Image 2.1 LoRA APIQwen Image 2.1 images and edits with up to three of your LoRAsFrom 5 credits per image
• Qwen Image 3.0 Pro APIFine typography plus lifelike skin, fabric and glassFrom 6 credits per image
• Qwen Image 3.0 APIReadable signs and headlines rendered into the imageFrom 5 credits per image
• Qwen Image 2 APISimple prompt-to-image and one-photo editsFrom 6 credits per image
• Qwen Image 2512 LoRA APIQwen Image 2512 renders and edits with your own LoRAsFrom 5 credits per image
• Qwen Image Edit 2509 LoRA APIMulti-photo Qwen edits that load your own LoRAsFrom 5 credits per image
• Qwen Image 1.0 LoRA APIOriginal Qwen-Image with clean lettering and your own LoRAsFrom 5 credits per image

Other image APIs

• Z-Image Base LoRA APIZ-Image Base with up to three of your own LoRAsFrom 3 credits per image
• MiniMax H3 Image LoRA APIH3 stills and edits with up to three of your LoRAsFrom 7 credits per image
• Anima Turbo LoRA APIAnime art with your own style or character LoRAsFrom 3 credits per image
• Krea 2 Turbo LoRA APIKrea 2 Turbo images and restyles with your own LoRAsFrom 4 credits per image
• P-Image LoRA APIFast P-Image renders and edits with one LoRA of your ownFrom 1 credit per image
• FLUX.2 Klein 9B LoRA APIFast FLUX.2 Klein 9B that loads up to three of your LoRAsFrom 3 credits per image

More about Qwen Image 2.1 LoRA

Try it

Qwen Image 2.1 LoRA AI Image Generator onlineCreate images with Qwen Image 2.1 LoRA in PopNow

Docs

Qwen Image 2.1 LoRA model ID and input schema in the API docsPopNow API quickstart: your first request in five minutesImage generation guide for the PopNow APIWebhooks: get Qwen Image 2.1 LoRA results without polling