Model overview
Seedream 5.0 Flash Layer Splitting splits a single image into one base layer and up to 16 transparent layers, and returns the layer hierarchy and bounding boxes. Resolution options include auto, 1K, 1.5K, and 2K; pricing is based on the actual number of output images, with the same unit price across all tiers.
The summary below is compiled from sources by an AI model, not our own testing. Verify selection decisions against original sources and your workload.
Key features
- Supports decomposing a single input image into 1 base image and up to 16 separate layers with transparent channels.
- The output includes the z-order and precise bounding box information for each layer.
- Resolution options include auto, 1K, 1.5K, and 2K, catering to varying precision requirements.
Limitations & considerations
- Only image input is accepted; plain text is not supported for generating layer structures.
- The maximum number of layers is 16; any layers beyond this limit cannot be decomposed.
- Billing is calculated in tiers based on the actual number of output layers (including the base layer), and bulk discounts or subscription plans are not supported.
Source prices & cost analysis
Prices vary by channel, region and configuration. Channel quotes are listed independently and are not treated as a single official provider price.
| Source / channel | Input | Output | Billing unit | Collected at |
|---|---|---|---|---|
| Apilio · default ↗ | Not verified | Not verified | per output image | 2026/11/2 06:25:12 |
Estimate monthly base token costs
Estimates exclude cache discounts, tiered pricing, multimodal surcharges, web search, taxes and additional reasoning output.
Public benchmarks & run configuration
暂无已采集的可核验公开评分。模型未上榜不代表能力较低。
View full indexed rankings →Frequently asked questions
Does this model support a custom number of layers?
Not supported; the number of layers is automatically inferred by the model, with a maximum of 16 transparent layers plus one background layer, and cannot be manually specified.
What is the coordinate format of the output bounding box?
The documentation does not specify the exact coordinate format of the bounding box (e.g., relative normalized values or absolute pixel values), nor does it mention the origin of the coordinate system.
How is the auto resolution specifically determined?
The documentation does not define the decision logic for the 'auto' resolution, such as whether it is based on input dimensions, aspect ratio, or content complexity.
Do open weights permit free commercial use?
Not necessarily. Check the version license, commercial conditions and usage policies. Self-hosting incurs compute and maintenance costs.
Are these real-time billing prices?
These are timestamped source snapshots, not live billing prices. Check the channel for current versions, regions, discounts and billing methods.
Sources & traceability
许可证：来源未披露，须进一步核验。模型介绍与第三方评测均以原来源的具体版本和方法为准。