Seedream 4.5 is ByteDance's 2K and 4K image model that improves on Seedream 4.0 with stronger reference fidelity in multi-image editing and better typography for dense, small text. On APIMart you call it as seedream-4.5, billed per image.
Service Highlights

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-4-5 | 2K/4K | 0.26Credits(~$0.026)/pic | 0.247Credits(~$0.0247)/pic | 0.234Credits(~$0.0234)/pic | 0.325Credits(~$0.0325)/pic | 20% |
* Actual costs are subject to final output.
Seedream 4.5 is the ByteDance Seedream model built for professional creatives: it keeps faces, lighting and colour from your reference images when it edits, and it lays out headlines and small print the way a designer would. Through one APIMart key you can generate at 2K or 4K, edit with several references and return up to 15 images in one task, at one price per image.
50K+
Active Users
99.9%
Uptime
2x
Faster
20%
Cost Savings
Call seedream-4.5 on APIMart with one POST request. Generation is asynchronous: submit a task, then poll the task endpoint to get the image URLs.
Save the key you created above in the APIMART_API_KEY environment variable and send it as Authorization: Bearer <key> on every request.
POST /v1/images/generations with model seedream-4.5, your prompt, size and resolution. Add image_urls to edit images or to generate a batch. The response returns a task_id.
Send GET /v1/tasks/{task_id} every few seconds until the task is completed, then read the image URLs from result.images[0].url. Results are kept for 72 hours, so download the files you want to keep.
See the Seedream 4.5 API reference for the full pixel table and every parameter.
curl -X POST https://api.apimart.ai/v1/images/generations \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedream-4.5",
"prompt": "Minimalist art fair poster on an off-white background, black line portrait slightly right of center, the title \"A/W ART FAIR\" in a serif font at the top left, three lines of small text at the bottom left with dates and venue",
"size": "3:4",
"resolution": "2K"
}'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-4.5",
"prompt": "Minimalist art fair poster on an off-white background, black line portrait slightly right of center, the title \"A/W ART FAIR\" in a serif font at the top left, three lines of small text at the bottom left with dates and venue",
"size": "3:4",
"resolution": "2K"
}
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; continue GET polling with this task ID")curl -X POST https://api.apimart.ai/v1/images/generations \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedream-4.5",
"prompt": "Dress the model in image 1 in the jacket from image 2, keep her pose, face and the studio lighting unchanged",
"image_urls": [
"https://example.com/model.jpg",
"https://example.com/jacket.jpg"
],
"size": "auto",
"resolution": "2K"
}'TASK_ID="task_xxx" # replace with the task_id returned by the POST request
curl "https://api.apimart.ai/v1/tasks/$TASK_ID" \
-H "Authorization: Bearer $APIMART_API_KEY"| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
model | string | Required | Default: seedream-4.5 | Model ID. seedream-4.5, seedream-4-5 and Seedream-4.5 are all accepted. |
prompt | string | Required | Default: — | Text description of the image or the edit. |
size | string | Optional | Default: 1:1 | Aspect ratio: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 2:1, 1:2, 21:9, 9:21, or auto to match the reference image (requires image_urls). |
resolution | string | Optional | Default: 2K | 2K or 4K, both at the same price. 1K is not supported. |
n | integer | Optional | Default: 1 | Number of images, 1–15. Values above 1 need at least one reference image, and references plus n must total 15 or fewer. |
image_urls | array | Optional | Default: — | Reference images as public URLs or base64 data URIs, JPEG or PNG, up to 10 MB each. Adding them switches to image editing or reference-based generation. |
optimize_prompt_options.mode | string | Optional | Default: standard | standard for higher quality, or fast for quicker results with regular quality. |
sequential_image_generation | string | Optional | Default: disabled | Set to auto with sequential_image_generation_options.max_images (1–15) to generate a related set. Becomes auto automatically when n is above 1. |
watermark | boolean | Optional | Default: false | Add a watermark to the generated images. |
nsfw_check | boolean | Optional | Default: false | Moderate the prompt and input images before the task is submitted. |
Compare every Seedream model on APIMart. One API key calls all of them through the same endpoint, billed per image.
| Model | Best for | Resolution |
|---|---|---|
| Seedream 5.0 Pro API | Quality-first single images with dense text, up to 10 references, point and box edits and layer splitting | 1K / 1.5K / 2K |
| Seedream 5.0 Flash API | The Seedream 5.0 Pro feature set at one lower price for every size, with no reference surcharge | 1K / 1.5K / 2K |
| Seedream 5.0 Lite API | 2K to 4K output, PNG or JPEG, and batches of up to 15 images per request | 2K / 3K / 4K |
| Seedream 4.5 APIThis page | 2K or 4K posters with stronger typography and reference fidelity, batches of up to 15 | 2K / 4K |
| Seedream 4.0 API | 1K, 2K and 4K at one price, image editing and batches of up to 15 with references | 1K / 2K / 4K |
Guides, comparisons and pricing analysis from the APIMart blog.
What changed from Seedream 4.0 and how to use it
Workflows that need readable text and consistent references
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-4.5, a prompt, size and resolution, then poll the returned task_id for the image URL.
What developers say about Seedream 4.5 API on APIMart
“The text rendering in Seedream 4.5 is remarkable. We generate promotional banners with accurate headlines and CTAs - something we couldn't achieve reliably with other models.”
Alex Thompson
Marketing Tech Lead
“Batch generation of 15 images per call transformed our workflow. We create entire product catalog variations in minutes instead of hours.”
Jessica Park
E-commerce Developer
“Reference image support keeps our brand consistent. Upload our style guide assets, and every generated image stays on-brand automatically.”
Ryan Chen
Creative Director
“4K output quality exceeded our expectations. We're using Seedream 4.5 for print materials that require high resolution without any upscaling artifacts.”
Maria Santos
Design Studio Owner
“The per-image price for this quality level is unbeatable. We migrated from a monthly subscription platform and now spend less while generating more content.”
Kevin Wu
Startup CTO
“The unified edit and generate capability means one API for everything. Simplified our entire image pipeline by removing three other services.”
Emma Liu
Backend Engineer
Answers about parameters, editing, resolution and pricing
Seedream 4.5 is ByteDance's image generation and editing model that scales up Seedream 4.0 for professional creative work: better reference fidelity in multi-image editing and stronger typography for dense text. On APIMart it is available through the Seedream 4.5 API with the model ID seedream-4.5.
It is an asynchronous REST API. You POST /v1/images/generations with model seedream-4.5 and a prompt, receive a task_id, then poll GET /v1/tasks/{task_id} until the task is completed and read the image URLs. The example section on this page has ready-to-run cURL and Python code.
Seedream 4.5 keeps the same request format but improves how it preserves the subject, lighting and colour of reference images and how it renders dense and small text. Seedream 4.5 outputs at 2K or 4K only, while Seedream 4.0 also offers 1K. Both support batches of up to 15 images with references.
Required: model and prompt. Optional: size (an aspect ratio or auto), resolution (2K or 4K), n (1–15), image_urls, optimize_prompt_options.mode (standard or fast), sequential_image_generation with sequential_image_generation_options.max_images, watermark and nsfw_check. The parameter table above lists types and defaults.
Yes. Set resolution to 4K, for example 4096x4096 for 1:1 or 5404x3040 for 16:9. 2K is the default. 2K and 4K cost the same per image, and 1K is not supported on Seedream 4.5.
Yes. Put the image in image_urls and describe the edit, for example "change the word in the red box to blue italic text" or "translate the image into Chinese, keep everything else unchanged". Use size auto to keep the original aspect ratio.
Reference images plus the number of generated images (n) must total 15 or fewer, so a single-image edit can use up to 14 references. Each reference must be a public URL or base64 data URI in JPEG or PNG, up to 10 MB and 6000x6000 pixels in total.
Provide at least one reference image and set n, or set sequential_image_generation to auto with max_images. A text-only request without image_urls returns a single image.
Seedream 4.5 is billed per generated image, and 2K and 4K cost the same. The pricing table on this page shows the current Gold member price (20% below list) alongside Platinum (24%) and Diamond (28%) prices. Batch tasks are settled on the images actually produced.
No. APIMart has no free quota for the Seedream 4.5 API. It is pay-as-you-go with no subscription, and failed tasks are not charged.
Sign up for APIMart, open the API Keys page in your dashboard and create a key. The same key works for the Seedream 4.5 API and every other Seedream version and model on APIMart.
No. ByteDance offers Seedream 4.5 as a hosted model through its own apps and APIs and has not published model weights for download. You use it by sending requests to an API such as the Seedream 4.5 API on APIMart.
Pick Seedream 4.5 for 4K posters and multi-reference edits with the 4.x request format. Pick Seedream 5.0 Lite for 2K, 3K or 4K with PNG output, or Seedream 5.0 Pro and Flash for single 1K–2K images with layer splitting. The Seedream API overview compares all versions with live prices.
Describe the subject, layout, style and lighting in plain sentences and put on-image text in double quotes with its position, for example "title at the top left in a serif font". The Seedream 4.5 prompt guide on our blog has worked examples.
Yes. The playground at the top of this page runs real Seedream 4.5 generations in your browser. Sign in, add balance and click Run; each image is billed at the same price as an API call, with no subscription. Results are kept for 72 hours.
The full Seedream 4.5 reference, with every parameter, the pixel table for 2K and 4K and request examples, is in the APIMart docs under Images > Seedream 4.5. The task status endpoint is documented under Tasks.
Explore more models in the same category.

Seedream 4.0
Seedream 4.0 API is text-to-image and image editing model. Generate 4K images with industry-leading text rendering, batch processing, and multi-reference support.

Flux 2 Pro
Flux 2 delivers state-of-the-art AI image generation and editing. Create photorealistic images from text with Flux 2 Pro, edit existing images with natural language using Flux 2 Flex, and maintain style consistency with multi-reference support.

Flux 2 Flex
Flux 2 delivers state-of-the-art AI image generation and editing. Create photorealistic images from text with Flux 2 Pro, edit existing images with natural language using Flux 2 Flex, and maintain style consistency with multi-reference support.

Nano banana pro
Nano Banana 2 API is powered by Google's Gemini 3 Pro Image Preview- the most advanced AI image generation model with 4K native output, multi-step planning workflow, and built-in quality analysis. Features 340% faster processing and professional-grade editing capabilities.