How much does the Seedream 4.5 API cost?

On Wizzx the Seedream 4.5 API starts at $0.04/image (40 credits). Each tier above lists its parameters and billing unit, and failed tasks are refunded.

What is the Seedream 4.5 API model ID?

seedream-4.5. Submit tasks to POST /api/v1/task/submit/seedream-4.5 with your API key.

Am I charged if a task fails?

No. Credits are reserved when you submit a task and refunded if it fails. You are only charged for successful tasks.

Is this the official Seedream 4.5 API?

No. Seedream 4.5 models are developed by ByteDance. Wizzx is an independent API platform that gives you access to them with one API key, an asynchronous task API and pay-as-you-go credits.

Does the submit response contain the generated file?

Generation is asynchronous. The submit response identifies the task. Query its status until it finishes, then use the output URLs in the result. See the API reference for response fields and error handling.

ByteDance

Available via APIImage Generation

Seedream 4.5 API

Name: Seedream 4.5 API
Author: ByteDance

Edit across up to 14 references with ByteDance Seedream 4.5, which keeps reference detail and renders dense text and typography more reliably than 4.0.

Model ID: seedream-4.5
Starts at: $0.04/image
Free trial: $0.20 in credits

View API Documentation

New accounts start with $0.20 in free credits, about 5 images at the starting price.

Compare all Seedream API versions

Seedream 4.5 is developed by ByteDance. Wizzx provides API access to it and does not own the model. All trademarks belong to their respective owners.

Next: turn the image into a video with

Model Guide

About Seedream 4.5

Edit across up to 14 references with ByteDance Seedream 4.5, which keeps reference detail and renders dense text and typography more reliably than 4.0.

Why choose Seedream 4.5?

Seedream 4.5 is the scaled-up successor to Seedream 4.0. According to ByteDance Seed, the larger model accurately identifies the main subjects in multi-image editing, strictly preserves the details of reference images, and further improves typography and dense text rendering, so it can deliver professional creatives with high consistency and fidelity. On ByteDance's MagicBench evaluation, 4.5 improves on 4.0 in prompt adherence, alignment with the source image, and aesthetics.

Choose 4.5 over 4.0 when the result has to match its references closely or carry a lot of small text. Choose 4.0 when you want 1K drafts or broad knowledge-style exploration. In Wizzx both models share the same controls:

Wizzx control	Current behavior
Prompt	Required
Reference images	Optional, up to 14 images for fusion or editing; references do not change the price
Aspect ratio	1:1 (default), 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 21:9
Resolution	1K, 2K (default), or 4K; resolution does not change the per-image rate
Number of images	1 (default) to 15 per request; reference images plus requested images must not exceed 15
Watermark	Off by default; switch on to add the upstream AI-generated mark
Seed	Optional integer for more reproducible iterations

Suitable workflows

Multi-reference edits where the model must pick the right subject, such as keeping only one marked person from a group shot.
Material and styling changes that must keep pose, silhouette, and composition, like turning a metallic dress into clear water.
E-commerce and beauty key visuals with titles, prices, and sectioned ingredient or feature copy.
Translating or restyling the text inside an existing design while leaving the rest of the layout intact.
Posters, invitations, and product detail pages where small captions must stay legible.
Campaign sets that reuse the same product or character references across several outputs.

Prompting approach

Name the main subject explicitly and say where it is, especially when several references contain similar people or products; marking a region in the source image and referring to it ("the character in the green outline") also works. Separate what changes from what stays, such as "keep the pose and the flowing shape of the dress; change only the material". For text-heavy layouts, quote every string, give its position and relative size, and list sections in reading order. Avoid conflicting style instructions in one prompt, because 4.5 follows reference detail closely and contradictions show up in the result.

Wizzx scope

Wizzx charges per generated image at a flat rate: resolution, aspect ratio, watermark, and the number of reference images do not change it, while Number of Images multiplies it. Number of Images is the upper bound for the request, and credits are settled for the images actually delivered. The live price is shown in the pricing section on this page and comes from the backend catalog.

BytePlus documents 2K and 4K as Seedream 4.5's resolution levels (4.0 also lists 1K), so use 2K or 4K for final work with this model. ByteDance's API also accepts exact pixel sizes and adds a watermark by default; Wizzx currently exposes only the eight preset aspect ratios and resolution tiers, does not offer custom pixel dimensions or an output format choice, and turns the watermark off unless you enable it.

Source

Backend pricing

Seedream 4.5 API pricing

Model identity, billing unit and every price tier below are read from the backend catalog.

Per generated image

credits / image

= $0.04 USD / image

100 images ≈ $4.00 · 1,000 images ≈ $40.00

Official: ~~$0.045~~−11%

Seedream 4.5 API vs official pricing

Tier	Official from	Wizzx	Savings
Per generated image	$0.045	$0.04	−11%

1 credit = $0.001 USD. The final deduction returned by the API is authoritative.

Back to all models

Integrate the Seedream 4.5 API

Every request is asynchronous: submit the task, keep the task ID, then poll for the result. The body below uses the model's default settings.

1
Get a Seedream 4.5 API key
Sign in and create a key in the dashboard. Set WIZZX_API_KEY in your server environment; never expose it in browser code.
2
Submit a task
Send a POST request to the model endpoint. A successful submission returns data.task_id; keep that ID before polling. Submission is not generation completion.
3
Retrieve the result
Replace YOUR_TASK_ID with the returned ID and query /api/v1/task/status. Wait for SUCCEEDED and read the output URLs. Stop on FAILED or REVIEW_REQUIRED; do not automatically resubmit an uncertain task.

Submit a tasksubmit.sh

curl -X POST \
  https://api.wizzx.ai/api/v1/task/submit/seedream-4.5 \
  -H "Authorization: Bearer $WIZZX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "aspect_ratio": "1:1",
  "resolution": "2K",
  "batch_size": 1,
  "watermark": false,
  "prompt": "A ceramic coffee cup on a wooden table, soft morning light, product photography"
}'

Submit a tasksubmit.ts

const response = await fetch("https://api.wizzx.ai/api/v1/task/submit/seedream-4.5", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.WIZZX_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "aspect_ratio": "1:1",
    "resolution": "2K",
    "batch_size": 1,
    "watermark": false,
    "prompt": "A ceramic coffee cup on a wooden table, soft morning light, product photography"
  }),
});

const payload = await response.json();
const task = payload.data ?? payload;
const taskId = task.task_id;

Submit a tasksubmit.py

import os
import requests

response = requests.post(
    "https://api.wizzx.ai/api/v1/task/submit/seedream-4.5",
    headers={
        "Authorization": f"Bearer {os.environ['WIZZX_API_KEY']}",
    },
    json={
        "aspect_ratio": "1:1",
        "resolution": "2K",
        "batch_size": 1,
        "watermark": False,
        "prompt": "A ceramic coffee cup on a wooden table, soft morning light, product photography"
    },
)
payload = response.json()
task = payload.get("data", payload)
task_id = task["task_id"]

Retrieve the resultstatus.sh

curl -X POST \
  https://api.wizzx.ai/api/v1/task/status \
  -H "Authorization: Bearer $WIZZX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "task_id": "YOUR_TASK_ID"
}'

Retrieve the resultstatus.ts

const response = await fetch("https://api.wizzx.ai/api/v1/task/status", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.WIZZX_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ task_id: "YOUR_TASK_ID" }),
});

const payload = await response.json();
const task = payload.data ?? payload;

Retrieve the resultstatus.py

import os
import requests

response = requests.post(
    "https://api.wizzx.ai/api/v1/task/status",
    headers={
        "Authorization": f"Bearer {os.environ['WIZZX_API_KEY']}",
    },
    json={"task_id": "YOUR_TASK_ID"},
)
payload = response.json()
task = payload.get("data", payload)

API documentation Compare models

Seedream 4.5 API parameters

The fields this endpoint accepts on Wizzx, read from the same schema as the Playground form.

Parameter	Type	Default	Allowed values	Description
`prompt` Promptrequired	string	—	—	Describe the image you want to generate
`image_urls` Reference Images	url[] ≤ 14	—	—	Up to 14 reference images for fusion
`aspect_ratio` Aspect Ratio	string	1:1	1:1 · 4:3 · 3:4 · 16:9 · 9:16 · 3:2 · 2:3 · 21:9	—
`resolution` Resolution	string	2K	1K · 2K · 4K	Choose output detail. Resolution does not change the per-image rate.
`batch_size` Number of Images	integer	1	1 – 15	Generate 1 to 15 variations in this request.
`watermark` Watermark	boolean	false	—	—
`seed` Seed	integer	—	—	Random seed for reproducibility

API integration questions

How much does the Seedream 4.5 API cost?: On Wizzx the Seedream 4.5 API starts at $0.04/image (40 credits). Each tier above lists its parameters and billing unit, and failed tasks are refunded.
What is the Seedream 4.5 API model ID?: seedream-4.5. Submit tasks to POST /api/v1/task/submit/seedream-4.5 with your API key.
Am I charged if a task fails?: No. Credits are reserved when you submit a task and refunded if it fails. You are only charged for successful tasks.
Can I try the Seedream 4.5 API for free?: Yes. New accounts start with $0.20 in free credits, about 5 images at the starting price.
Is this the official Seedream 4.5 API?: No. Seedream 4.5 models are developed by ByteDance. Wizzx is an independent API platform that gives you access to them with one API key, an asynchronous task API and pay-as-you-go credits.
Does the submit response contain the generated file?: Generation is asynchronous. The submit response identifies the task. Query its status until it finishes, then use the output URLs in the result. See the API reference for response fields and error handling.

Wizzx API

Start building with the Seedream 4.5 API

New accounts get $0.20 in free credits. Pay as you go with no subscription, and failed tasks are refunded.

Try in Playground

Seedream 4.5 API

About Seedream 4.5

Why choose Seedream 4.5?

Suitable workflows

Prompting approach

Wizzx scope

Source

Seedream 4.5 API pricing

Per generated image

Seedream 4.5 API vs official pricing

Integrate the Seedream 4.5 API

Get a Seedream 4.5 API key

Submit a task

Retrieve the result

Seedream 4.5 API parameters

API integration questions

Seedream 4.0

Seedream 5.0 Pro

Qwen Image 2.0

Qwen Image 2.0 Pro

Start building with the Seedream 4.5 API

Seedream 4.5 API

About Seedream 4.5

Why choose Seedream 4.5?

Suitable workflows

Prompting approach

Wizzx scope

Source

Seedream 4.5 API pricing

Per generated image

Seedream 4.5 API vs official pricing

Integrate the Seedream 4.5 API

Get a Seedream 4.5 API key

Submit a task

Retrieve the result

Seedream 4.5 API parameters

API integration questions

Compare related models

Seedream 4.0

Seedream 5.0 Pro

Qwen Image 2.0

Qwen Image 2.0 Pro

Start building with the Seedream 4.5 API