Seedream 5.0 Pro is ByteDance's quality-first image model, served on APIMart as seedream-5-0-pro. One endpoint covers text to image, image editing with up to 10 reference images, point and box precision edits, and layer decomposition, at 1K, 1.5K, 2K or an exact pixel size.
Precise editing: add point or bbox coordinate tags to the prompt.

Image link valid for 72 hours
Transparent pricing with no hidden fees. Pay only for what you use.
| Variant | Tier | GoldCurrent price | Platinum | Diamond | Official | Your savings |
|---|---|---|---|---|---|---|
seedream-5-0-pro | 1K | 0.2925Credits(~$0.0293)/pic | 0.2779Credits(~$0.0278)/pic | 0.2633Credits(~$0.0263)/pic | 0.3656Credits(~$0.0366)/pic | 20% |
| 1.5K | 0.2925Credits(~$0.0293)/pic | 0.2779Credits(~$0.0278)/pic | 0.2633Credits(~$0.0263)/pic | 0.3656Credits(~$0.0366)/pic | 20% | |
| 2K | 0.585Credits(~$0.0585)/pic | 0.5558Credits(~$0.0556)/pic | 0.5265Credits(~$0.0527)/pic | 0.7313Credits(~$0.0731)/pic | 20% | |
| Reference image (first 1 free, then each) | 0.0195Credits(~$0.002)/pic | 0.0185Credits(~$0.0019)/pic | 0.0176Credits(~$0.0018)/pic | 0.0244Credits(~$0.0024)/pic | 20% |
| Variant | Tier | GoldCurrent price | Platinum | Diamond | Official | Your savings |
|---|---|---|---|---|---|---|
seedream-5-0-pro | 1K | 0.1463Credits(~$0.0146)/pic | 0.1389Credits(~$0.0139)/pic | 0.1316Credits(~$0.0132)/pic | 0.1828Credits(~$0.0183)/pic | 20% |
| 2K | 0.2925Credits(~$0.0293)/pic | 0.2779Credits(~$0.0278)/pic | 0.2633Credits(~$0.0263)/pic | 0.3656Credits(~$0.0366)/pic | 20% |
* Actual costs are subject to final output.
The Seedream 5.0 Pro API gives you ByteDance's quality-first image model through one APIMart key. Render posters with readable headlines, keep product shots consistent across a catalog, and edit photos with up to 10 reference images. Output is billed per image: 1K and 1.5K share one price, 2K costs more, and every reference image after the first is charged separately.
50K+
Active Users
99.9%
Uptime
2x
Faster
20%
Cost Savings
Comparing every Seedream version, from 4.0 to 5.0 Pro, Lite and Flash? See the Seedream API overview
Call seedream-5-0-pro on APIMart with one POST request. Generation is asynchronous: submit a task, then poll the task endpoint to get the image URL.
Save the key you created above in the APIMART_API_KEY environment variable and send it as Authorization: Bearer <key> on every request. An invalid key returns HTTP 401 and an empty balance returns 402.
POST /v1/images/generations with model seedream-5-0-pro, your prompt, size and resolution. Add image_urls to edit images. The response returns a task_id.
Send GET /v1/tasks/{task_id} every few seconds until the task is completed, then read the image URL from result.images[0].url[0]. URLs expire at expires_at, so download the files you want to keep. Failed tasks are refunded in full.
See the Seedream 5.0 Pro API reference for every parameter, layer response field and error code.
curl -X POST https://api.apimart.ai/v1/images/generations \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedream-5-0-pro",
"prompt": "A minimalist summer sale poster with the headline \"SUMMER SALE\" in bold white letters, sliced citrus fruit on a pastel blue background",
"size": "16:9",
"resolution": "1.5K",
"output_format": "png"
}'import os
import time
import requests
API_BASE = "https://api.apimart.ai"
headers = {"Authorization": f"Bearer {os.environ['APIMART_API_KEY']}"}
# 1. Submit the generation task
payload = {
"model": "seedream-5-0-pro",
"prompt": "A minimalist summer sale poster with the headline \"SUMMER SALE\" in bold white letters, sliced citrus fruit on a pastel blue background",
"size": "16:9",
"resolution": "1.5K",
"output_format": "png"
}
resp = requests.post(f"{API_BASE}/v1/images/generations", headers=headers, json=payload, timeout=60)
resp.raise_for_status()
task_id = resp.json()["data"][0]["task_id"]
# 2. Poll the task until it finishes
for _ in range(200):
task_resp = requests.get(f"{API_BASE}/v1/tasks/{task_id}", headers=headers, timeout=30)
task_resp.raise_for_status()
task = task_resp.json()["data"]
if task["status"] == "completed":
print(task["result"]["images"][0]["url"][0])
break
if task["status"] == "failed":
raise RuntimeError(task.get("error", {}).get("message", "generation failed"))
time.sleep(3)
else:
raise TimeoutError(f"task {task_id} did not finish in time")curl -X POST https://api.apimart.ai/v1/images/generations \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedream-5-0-pro",
"prompt": "Dress the person in image 1 in the dress from image 2, keep the pose, face and lighting unchanged",
"image_urls": [
"https://example.com/person.jpg",
"https://example.com/dress.jpg"
],
"resolution": "2K",
"size": "auto"
}'TASK_ID="task_xxx" # replace with the task_id returned by the POST request
curl "https://api.apimart.ai/v1/tasks/$TASK_ID" \
-H "Authorization: Bearer $APIMART_API_KEY"| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
model | string | Required | Default: seedream-5-0-pro | Model ID. Use seedream-5-0-pro; seedream-5.0-pro is also accepted. |
prompt | string | Required | Default: — | Text description of the image or the edit. Keep it under about 600 English words. May be omitted when layer_decomposition is true. |
size | string | Optional | Default: auto | A tier (1K, 1.5K, 2K), auto, an aspect ratio (1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 2:1, 1:2, 21:9) or an exact pixel size such as 2048x1024 (921,600–4,624,220 total pixels, ratio 1:16 to 16:1). |
resolution | string | Optional | Default: 1K | Resolution tier: 1K, 1.5K or 2K. 1.5K costs the same as 1K. Ignored when size is an exact pixel size. |
image_urls | array | Optional | Default: — | Up to 10 reference images as public URLs or base64 data URIs, max 30 MB each. Adding them switches to image editing. The first image is free; each additional image adds a fee. |
output_format | string | Optional | Default: jpeg | Output format: jpeg or png. |
background | string | Optional | Default: opaque | opaque or transparent. transparent requires exactly one input image with an alpha channel and output_format png. |
layer_decomposition | boolean | Optional | Default: false | Split exactly one PNG or JPEG input into a base image and up to 16 transparent PNG layers. size then accepts only 1K, 1.5K, 2K or auto. |
optimize_prompt_options | object | Optional | Default: {"mode": "standard"} | Prompt optimization. Only the standard mode is supported. |
watermark | boolean | Optional | Default: false | Add an "AI generated" watermark in the bottom-right corner. |
nsfw_check | boolean | Optional | Default: false | Moderate the prompt and input images before the task is submitted. |
n | integer | Optional | Default: 1 | Number of images. Only 1 is supported; for batches use Seedream 5.0 Lite. |
Compare every Seedream model on APIMart. One API key calls all of them, billed per image.
| Model | Best for | Resolution |
|---|---|---|
| Seedream 5.0 Pro APIThis page | Quality-first cinematic images, precise text, up to 10 references and layer splitting | 1K / 1.5K / 2K |
| Seedream 5.0 Flash API | Fast cinematic images with precise text and up to 10 references | 1K / 1.5K / 2K |
| Seedream 5.0 Lite API | 4K images with advanced text rendering and batches of up to 15 | 2K / 3K / 4K |
| Seedream 4.5 API | 4K text-to-image with precise text rendering and multi-reference batches | 2K / 4K |
| Seedream 4.0 API | 4K text-to-image and image editing with batch output | 1K / 2K / 4K |
Guides, comparisons and pricing analysis from the APIMart blog.
Generation, editing and layer splitting on one endpoint, all from the same model ID
Three workflows the model handles best: text posters, product images and multi-reference editing
Set up your account, add balance and send your first request
Sign up for APIMart and top up your balance. Billing is pay-as-you-go with no minimum, and the balance works across every model on the platform.
Create a key in the console. You will send it in the Authorization header of every request.
POST to /v1/images/generations with model seedream-5-0-pro, a prompt and a resolution or exact size, then poll the returned task_id for the image URL.
What developers say about Seedream 5.0 Pro on APIMart
“The cinematic quality is on another level. Our hero banners look like they came out of a photo studio, with lighting and detail we couldn't get elsewhere.”
Alex Thompson
Marketing Tech Lead
“Text rendering is spot on. We generate promotional posters with accurate headlines and CTAs directly from prompts — no more manual typesetting.”
Jessica Park
E-commerce Developer
“Multi-reference fusion made our virtual try-on feature possible. Combining a model and a garment from two references just works, and it is now core to our product.”
Ryan Chen
Creative Director
“1.5K and 2K output exceeded our expectations. We use it for editorial imagery and need no upscaling to hit high resolution.”
Maria Santos
Design Studio Owner
“Pay-as-you-go pricing works for us. We get premium visuals with no monthly commitment or hidden fees.”
Kevin Wu
Startup CTO
“Unified editing and generation means one service for everything. It replaced three other tools in our pipeline.”
Emma Liu
Backend Engineer
Answers about parameters, editing, resolution and pricing
Seedream 5.0 Pro is the quality-first image generation and editing model in ByteDance's Seedream 5.0 family. On APIMart its model ID is seedream-5-0-pro (seedream-5.0-pro is also accepted), and it is listed upstream as doubao-seedream-5-0-pro. You may also see it written as Seedream 5 Pro, Seedream V5 Pro or ByteDance Seedream 5.0 Pro.
It is an asynchronous REST API. You POST /v1/images/generations with model seedream-5-0-pro and a prompt, receive a task_id, then poll GET /v1/tasks/{task_id} until the task is completed and read the image URL. The example section on this page has copy-ready cURL and Python code.
Required: model and prompt (prompt may be omitted only for layer decomposition). Optional: size, resolution (1K, 1.5K or 2K), image_urls (up to 10), output_format (jpeg or png), background (opaque or transparent), layer_decomposition, optimize_prompt_options, watermark and nsfw_check. n must be 1. The parameter table above lists defaults and limits.
Use a 1K, 1.5K or 2K tier with auto or one of ten aspect ratios (1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 2:1, 1:2, 21:9), or an exact pixel size between 921,600 and 4,624,220 total pixels with an aspect ratio from 1:16 to 16:1. 1.5K costs the same as 1K. Exact sizes are billed by output area: up to 2,601,124 pixels uses the lower tier, larger outputs use the 2K tier.
Yes. Add one or more images in image_urls and describe the edit in the prompt. You can target a region with <point> or <bbox> coordinates or with hand-drawn marks on the image, and you can keep a transparent background by sending one PNG with an alpha channel together with background set to transparent and output_format set to png.
Send up to 10 images in image_urls as public URLs or base64 data URIs (jpeg, png, webp, bmp, tiff, gif, heic or heif, up to 30 MB and 36 million pixels each). The first reference image is free and every additional reference adds a fixed per-image fee, listed in the pricing table on this page.
With layer_decomposition set to true and exactly one PNG or JPEG input, the model returns a base image plus up to 16 transparent PNG layers with names and bounding boxes. Each output image is billed by its own size. The request reserves the cost of up to 17 images when submitted, then refunds the difference once the actual layers are counted.
Choose Seedream 5.0 Pro for single images where quality, layer splitting and paid multi-reference editing matter. Choose Seedream 5.0 Lite when you need 2K to 4K output or batches of up to 15 images per request. Choose Seedream 5.0 Flash for high volume at a lower per-image price, with one price for every size. The comparison table below links each model page and the full Seedream API overview.
Describe the subject, style, lighting and layout in plain sentences and keep the prompt under about 600 English words, since very long prompts can lose detail. Put on-image text in double quotes, refer to references as image 1, image 2 and so on, and use coordinates when an edit must stay in one place.
Authenticate with Authorization: Bearer <your key>, POST /v1/images/generations on https://api.apimart.ai, then poll GET /v1/tasks/{task_id}. The full reference, including layer decomposition responses and error codes, is at docs.apimart.ai under Seedream-5.0-Pro image generation.
Seedream 5.0 Pro pricing is per output image: 1K and 1.5K share one price, 2K is a higher tier, and each reference image after the first adds a fixed fee. The pricing table on this page shows current Gold member prices (20% off); Platinum and Diamond members save up to 28%, and you only pay for successful requests.
No. There is no free quota for the Seedream 5.0 Pro API on APIMart. Billing is pay-as-you-go, you only pay for successful requests, and every price comes with 20% off.
Sign up for an APIMart account, open the API Keys page and create a key. The same key works for the Seedream 5.0 Pro API and every other model on APIMart, with pay-as-you-go billing and no subscription.
No. It is a single-image model: n must be 1, and sequential_image_generation, stream and tools are rejected with HTTP 400 before any charge. For batches of up to 15 images in one request, use Seedream 5.0 Lite.
Explore more models in the same category.

Nano Banana 2 Lite
Nano Banana 2 Lite is the lightweight model in the Nano Banana 2 family, generating and editing images at 1K with one image per request and up to 14 reference images. Official channel.

Midjourney
Midjourney is a leading text-to-image AI model renowned for its painterly aesthetics, strong composition, and cinematic lighting. Generate high-quality images from text prompts or reference images, with fine-grained control over style, aspect ratio, and creative parameters — accessible programmatically through APIMart's unified API, no Discord required.

Wan 2.7 Image Pro
wan‑2‑7‑image is a powerful image generation model in Alibaba’s Wan 2.7 series that creates high‑quality visuals from text prompts. It produces detailed and realistic images with strong semantic understanding and refined composition. The model supports multimodal input—such as combining text with reference images—to influence style, layout, or visual direction. It’s well‑suited for creative workflows like product visuals, marketing graphics, social media content, and artistic illustrations, delivering both fidelity and expressive visual output.

Wan 2.7 Image
wan‑2‑7‑image is an advanced image generation model in Alibaba’s Wan 2.7 series designed to create high‑quality visuals from text prompts. It excels at generating detailed, realistic images with accurate object representation and rich composition. The model supports multimodal input (e.g., combining text with reference images) to influence style and structure, and it’s well‑suited for creative workflows such as marketing assets, product visuals, social media graphics, and artistic content production. With strong semantic understanding and prompt adherence, wan‑2‑7‑image delivers both fidelity and expressive visual output.