Access Seedream 5.5 through EvoLink's unified image API: 1K, 1.5K or 2K output with auto, transparent or opaque backgrounds and up to 10 reference inputs. Seedream 5.5 Layerize splits one image into a base image and up to 99 PNG layers with optional text recognition.
Test Seedream 5.5 output quality, transparent backgrounds and layer splitting, estimate credit usage, and generate a sample before you formally integrate.
Create image task
Fixed to doubao-seedream-5.5Text-to-image, image-to-image and editing. One image per request: 1K and 1.5K share one rate, 2K is billed higher, and each input image is billed.
Upload source or reference images. Each input image is billed at the input-image rate.
1K and 1.5K share one rate; 2K is billed at the higher rate.
Auto lets the model decide; the task result reports whether the image came back transparent or opaque.
Generated images will appear here.
Results are kept for 24 hours — download anything you want to keep.
Output is billed by tier - 1K and 1.5K share one rate, 2K is billed higher - and each input image is billed per image. Pick a testing budget, or expand for detailed rates.
First Test Cost
Seedream 5.5 - Text-to-Image - 1K - 1 imageOne image at the 1K or 1.5K tier, no input images
About 597 tests with $10 in credits.
In Playground, keep Seedream 5.5 selected and generate one 1K or 1.5K image without input images. 1.5K costs the same as 1K, so test at 1.5K; switch to 2K or add reference images only when the layout needs them.
Seedream 5.5 testing budget guide
Pick an amount based on how many tests you expect.Good for a first validation.
Good for prompt iteration.
Good before production integration.
Output images are billed by tier and input images per image; Layer Split bills the input image once plus every image produced.
doubao-seedream-5.5| Item | Rule | Rate | Billing |
|---|---|---|---|
| 1K / 1.5K output | Single generated image at the 1K or 1.5K tier, or a custom size up to 2,610,000 pixels. Both tiers cost the same, so 1.5K is the better pick. | $0.016/image-20% 1.088 cr/image$0.020 official price | Output image |
| 2K output | Single generated image at the 2K tier, or a custom size above 2,610,000 pixels. | $0.022/image-20% 1.4688 cr/image$0.027 official price | Output image |
doubao-seedream-5.5-layerize| Item | Rule | Rate | Billing |
|---|---|---|---|
| 1K / 1.5K output | Each image produced by Layer Split - the base image and every layer - whose size is at or below 2,610,000 pixels. | $0.016/image-20% 1.088 cr/image$0.020 official price | Output image |
| 2K output | Each image produced by Layer Split above 2,610,000 pixels. A run yields one base image plus up to 99 layers, each priced by its own pixel count. | $0.022/image-20% 1.4688 cr/image$0.027 official price | Output image |
| Item | Rule | Rate | Billing |
|---|---|---|---|
| Input images | Applied to each source or reference image in image_urls, up to 10 per request. Layer Split bills its single input image at the same rate. | $0.0064/image-20% 0.4352 cr/image$0.0080 official price | Input image |
| Layer Split reserve | One input image plus n output images at the 2K rate are reserved when the task starts (n defaults to 30). You pay only for the images actually produced, each at its own tier; the rest is refunded. | $0.655 · 44.4992 cr | Refunded after completion |
Use Seedream 5.5 for product shots, ad iterations and multi-reference compositions, with transparent backgrounds for cut-out assets. Ordinary generation returns one image; Layerize returns a base image and up to 99 separate transparent layers, with recognized text for clean text layers. Compare output quality on your own prompts before choosing a production workflow.
doubao-seedream-5.5Generate or edit a flat image from a prompt, with up to 10 reference images, ratio presets or custom pixel sizes, 1K/1.5K/2K quality tiers, and auto, transparent or opaque backgrounds.
doubao-seedream-5.5-layerizeSend exactly one PNG or JPEG and get a base image plus up to 99 transparent PNG layers with stacking order, bounding boxes and optional recognized text, billed per image produced.
These fields cover the main integration decisions. Open the API tab for the full request and response schema.
modelstringdoubao-seedream-5.5 for generation and editing; doubao-seedream-5.5-layerize for Layer Split.
promptstringDescribe the image. Up to 4000 tokens; under about 600 English words works best. Optional for Layer Split.
image_urlsarrayUp to 10 reference images for image-to-image or editing, each billed per image. Layer Split takes exactly one.
sizestringauto, an aspect ratio such as 1:1 or 16:9, or custom WIDTHxHEIGHT pixels.
qualityenum1K, 1.5K or 2K for auto or ratio sizes. 1K and 1.5K share one rate; 2K is billed higher.
backgroundenumauto lets the model decide and reports the result; transparent returns a real alpha background as PNG; opaque returns a solid background.
output_formatenumpng or jpeg. Transparent backgrounds require png.
nintegerLayer Split only: maximum number of output images including the base image, 2–100. Sets the reserve.
model_params.text_recognitionbooleanLayer Split only: return recognized text, lines and font styles for clean text layers.
Use the EvoLink API for product integration and batch jobs, or call it from Codex, Claude, or Gemini for fast creative and development workflows. Both paths share the same EvoLink API key, balance, model routes, and task history.
Best for: product backends, batch jobs, automated workflows
Call EvoLink’s unified image API from your server and control the model ID, parameters, task queue, callbacks, and result storage.
Best for: creative and development tasks in Codex, Claude, and Gemini
Give the Agent your output goal, assets, and acceptance criteria. It can choose the route, assemble the request, track the task, and return the result without requiring you to hand-code every step.
This example shows the shortest runnable flow: create a task, save the returned task ID, then poll or wait for an HTTPS callback. Open the API tab for the complete parameter and response reference.
curl -X POST https://api.evolink.ai/v1/images/generations \
-H "Authorization: Bearer $EVOLINK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedream-5.5",
"prompt": "A serene lake reflecting the sunset, editorial photography",
"size": "16:9",
"quality": "2K",
"background": "auto",
"output_format": "png",
"callback_url": "https://your-domain.com/webhooks/image-task-completed"
}'
# Save the returned task ID, then query:
curl https://api.evolink.ai/v1/tasks/{task_id} \
-H "Authorization: Bearer $EVOLINK_API_KEY"Check every value against the parameter cards above: ranges, allowed options, asset counts, and file types must match the selected route.
Check the Authorization bearer token and confirm the available balance and task billing in the console.
Review likeness rights, brand assets, readable text, sensitive content, and input-asset compliance.
A Seedream 5.5 task that reaches a final failed state is not charged. Keep the task ID, inspect the final state, and identify whether the cause is validation, review, or execution before retrying.
callback_url must be a public HTTPS URL, not localhost or a private network address. Always keep the task ID as a polling fallback.
| Feature | Seedream 5.5 | Nano Banana | GPT Image |
|---|---|---|---|
| Starting price | ~$0.016 / image | See selected model | Token or tier based |
| Output quality | 1K / 1.5K same rate, 2K higher | 1K / 2K / 4K by model | Quality tiers |
| Transparent output | Auto / transparent / opaque | Model dependent | Supported |
| Layer output | Base + up to 99 PNG layers, text recognized | No native layer output | No native layer output |
| Best for | Cut-out assets, references, layer workflows | High-quality image generation | OpenAI image workflows |
1K and 1.5K share one rate, 2K is billed higher, and every image in image_urls adds one input-image charge. Custom sizes are billed by their pixel count.
A Layer Split run returns one base image plus up to 99 layers, each billed at its own tier, plus one input-image charge. The reserve is one input plus n images at the 2K rate; the unused part is refunded.
Seedream 5.5 accepts prompt_priority: standard only. A Seedream 5.0 Pro request using fast must drop it before switching the model ID; review the new background default too.
See the tier rate, input-image charge and your estimate before you add credits.
One API key works across Seedream, Nano Banana, GPT Image, video, audio, and LLM models.
Track credits, spending, and model usage in one console.
See whether an image task is submitted, processing, completed, or failed.
Test drafts, transparent outputs, edits, and layer splits on one page before you integrate.
Compare output limits, reference handling, background options and billing across Seedream models under one EvoLink account and API key.
Seedream 5.0 Pro
Compare Pro for its standard and fast prompt modes and discounted resolution-based pricing. Evaluate output quality using the same prompts and references.
View API
Seedream 5.0 Flash
Compare Flash for flat per-image pricing with free reference images when speed and volume matter more than transparent output.
View API
Seedream 5.0 Lite
Evaluate Lite when your workflow needs web-search-enhanced generation. Check its own resolution, reference limits and optional search charges.
View API
GPT Image 2.5 Flare
OpenAI's default GPT Image 2.5 route for fast, high-quality everyday generation, billed by tokens.
View API
GPT Image 2.5 Sunburst
GPT Image 2.5 variant tuned for editing precision, trading longer generation times for finer detail control.
View API
Grok Imagine Image 2.0
xAI image model for text-to-image and editing with 1–3 reference images, with 1K/2K output.
View APISeedream 5.5 is ByteDance's newest image generation and editing model, adding transparent output and up to 99-layer decomposition. On EvoLink, use doubao-seedream-5.5 for ordinary generation or editing, and doubao-seedream-5.5-layerize for layer decomposition. The two models share this page but have different input and output rules.
They are separate models. doubao-seedream-5.5 generates or edits a flat image and accepts ratios, custom pixel sizes and the background parameter. doubao-seedream-5.5-layerize takes exactly one PNG or JPEG, accepts quality tiers only, and returns a base image plus up to 99 transparent PNG layers with stacking order, bounding boxes and, when enabled, recognized text.
Check the pricing section for the current EvoLink rate. Output images are billed by tier - 1K and 1.5K share one rate, 2K is billed higher - and each image_urls input adds an input-image charge. Layerize bills its input image once plus every image produced, each at its own tier.
The number of layers is unknown when the task starts, so one input image plus n output images at the 2K rate are reserved (n defaults to 30). You pay only for the images actually produced, each at its own tier, and the rest returns to your balance.
Pass background: transparent to get an image with a real alpha channel; output is always png. It works for text-to-image and for image-to-image with up to 10 references - no alpha-channel input is required. Pass auto to let the model decide, and read result_data[].background to see which it used.
No. Seedream 5.5 accepts prompt_priority: standard only; fast is rejected before billing. When migrating a Seedream 5.0 Pro request, remove fast and compare your own turnaround time rather than assuming a fixed generation speed.
Yes. Use the Playground to test output quality, estimate credits, and inspect the request JSON before wiring it into your product.
Not for ordinary generation: doubao-seedream-5.5 returns one image per request. For doubao-seedream-5.5-layerize, n caps the total number of output images including the base image (2–100, default 30).
Set model_params.text_recognition to true on a Layerize request. Clean text layers then include text_recognition in result_data: the full text, each line's bounding box, and each run's text with a similar font family, color and bold / italic / underline / strikethrough / shadow / translucent flags. Stylized or handwritten text is not recognized.
Failed tasks are not billed. Check the task status and error details in the response or console, adjust the request, and try again.
Use an EvoLink API key to POST /v1/images/generations with the Seedream 5.5 model ID and the fields shown in this page's API reference. Save the returned id and query /v1/tasks/{task_id}, or provide an HTTPS callback_url. The create response is a task record, not the finished image.
Evaluate Seedream 5.5 for transparent output, auto backgrounds and up to 99-layer decomposition with text recognition. Pro offers standard and fast prompt modes with discounted tier pricing. Flash offers one flat rate with free references. Compare the family page, then test the same prompts; price and tier names do not prove a quality ranking.
Seedream 5.5 supports 1K, 1.5K and 2K output tiers, not a 4K tier. For ordinary generation, size can be auto, an aspect ratio or WIDTHxHEIGHT within the documented pixel and ratio limits, billed by pixel count. Layerize uses quality: auto, 1K, 1.5K or 2K and does not accept aspect ratios or custom pixel sizes.