Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Alibaba Cloud / Wanx design images

Wanx Background Generation v2

Product-background generation from a transparent foreground, with text, reference-image and layered edge guidance. The documented API selector supports v2 and v3 processing versions.

historicalClosed weightsImage editing
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
33 of 41 specification fields collected
Output resolutionMatches the main input image; longest side at most 2,048 pixelsSee specification notes below
Output formatsNot collectedAwaiting source review
Input billingSee ratesAlibaba Model Studio · Beijing · historical
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected
Maximum input tokens
Not collected
Maximum output
Not collected
Input modalities
Text, ImageSource for Input modalities (opens in a new tab)

The API supports text guidance although the summary card lists only Image.

Reasoning
Not collected
Streaming
Not collected
Knowledge cutoff
Not collected
Parameters
Not collected
Architecture
Not collected
API aliases
wanx-background-generation-v2Source for API aliases (opens in a new tab)
Published availability regions
China (Beijing)Source for Published availability regions (opens in a new tab)
Request mode
Async HTTP submission with X-DashScope-Async: enable, followed by task-status queries. This is not Batch API support.Source for Request mode (opens in a new tab)
Published price
Historical CNY 0.08 per generated image.Source for Published price (opens in a new tab)
Usage metering
Only output images are billed; input is free. Historical listing, current paid access unverified.Source for Usage metering (opens in a new tab)
Free usage allowance
500 images. API/guide says 180 days after activation; central pricing says 90 days from latest activation, release or approval. Account entitlement unresolved.Source for Free usage allowance (opens in a new tab)

The conflicting 180-day term is preserved from the API and guide, not replaced by an invented common duration.

Required request fields
base_image_url plus at least one of ref_prompt or ref_image_url. Public HTTP(S) images.Source for Required request fields (opens in a new tab)
Image input limits
Main image: transparent four-channel RGBA PNG; longest side at most 2,048 pixels.Source for Image input limits (opens in a new tab)
Input image formats
RGBA PNG (main image), JPG, PNG, WEBP and other documented reference formatsSource for Input image formats (opens in a new tab)

The reference format list is non-exhaustive; main-image RGBA PNG requirement is strict.

Output resolutions
Matches the main input image; longest side at most 2,048 pixelsSource for Output resolutions (opens in a new tab)
Images per request
1–4 output images; default 1Source for Images per request (opens in a new tab)
Prompt limits
Per prompt: English up to 150 words; Chinese described as 100–120 characters. Excess is ignored. Applies to positive, negative and individual edge prompts. The exact Chinese cutoff remains ambiguous.Source for Prompt limits (opens in a new tab)
Reference image limits
Reference image may be RGB or RGBA. Fully transparent pixels are excluded from guidance. Reference-image size limits beyond the reviewed format rules remain unverified.Source for Reference image limits (opens in a new tab)
Layer and edge guidance
At most 10 foreground + background elements combined. RGBA layers resize to match the main image, retain bottom-to-top order and may occlude or sit behind the product. Optional prompt lists must match each layer list in length and order; empty strings preserve unprompted positions.Source for Layer and edge guidance (opens in a new tab)
Processing versions
v2: default, older/faster processing, v3: newer processing, provider reports higher quality and slower responseSource for Processing versions (opens in a new tab)

Both use model=wanx-background-generation-v2. Select parameters.model_version; wanx-background-generation-v3 is explicitly rejected as an API model name. Private checkpoint lineage and v3 release date remain unknown.

Generation controls
noise_level: 0–999, default 300, only with an image reference. ref_prompt_weight: 0–1, default 0.5, only when both reference text and image are provided.Source for Generation controls (opens in a new tab)
Deprecated controls
title, sub_title, scene_typeSource for Deprecated controls (opens in a new tab)

Deprecated text-overlay controls are not promoted as the current workflow.

Output link expiry
Task IDs and generated-image URLs are valid for 24 hours. Input retention and training use remain unknown.Source for Output link expiry (opens in a new tab)
Regional route evidence
Guide gives Beijing workspace URLs. API declares Beijing-only scope but uses dashscope-intl in submission examples and mentions Singapore for result queries. No complete Singapore access record is inferred.Source for Regional route evidence (opens in a new tab)

Contradictory API routes are retained in the API source.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Alibaba Model Studio · Beijing · historical

Provider source (opens in a new tab)
API model ID
wanx-background-generation-v2
Access
POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation. Async HTTP submission with X-DashScope-Async: enable, followed by task-status queries. This is not Batch API support. Historical guide route; other published routes conflict and current access is unverified.
Regions
China (Beijing)
Published request quota
120 requests per minuteSource for Published request quota (opens in a new tab)
Task submission rate
2 submissions per secondSource for Task submission rate (opens in a new tab)
Quota scope
Per-model limits shared by the Alibaba Cloud account and RAM usersSource for Quota scope (opens in a new tab)
Free usage allowance
500 images. API/guide says 180 days after activation; central pricing says 90 days from latest activation, release or approval. Account entitlement unresolved.Source for Free usage allowance (opens in a new tab)

The conflicting 180-day term is preserved from the API and guide, not replaced by an invented common duration.

Request mode
Async HTTP submission with X-DashScope-Async: enable, followed by task-status queries. This is not Batch API support.Source for Request mode (opens in a new tab)
Published token quota
3,000 tokens per minuteSource for Published token quota (opens in a new tab)

Not a context limit or a token-based image tariff.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Historical background generationGenerated image1CN¥0.08image

Rate conditions

  1. Published domestic output-only tariff; input is free. Current availability unresolved. Not a converted USD rate or text-question estimate.
Pricing source (opens in a new tab)

Data & enterprise terms

Result-link expiry is documented. Input retention, training use and account entitlement are not verified.

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Classification inference: separately named native Alibaba hosted release; reviewed official model cards and APIs describe service access but do not provide model weights. No exact released checkpoint is identified by those sources. Public Wan checkpoints, style controls and research architectures are not assumed equivalent.

Classification evidence (opens in a new tab)

Lifecycle dates

Model Studio release22 Mar 2024Source (opens in a new tab)

Sources reviewed