
Seedream
: How to use it, features, and the business problems it solves
Add bookmark
What is Seedream?
Seedream is an image generation and editing AI model family developed by ByteDance's Seed research team. The latest Seedream 5.0 excels at precise editing including layer separation and multilingual text rendering. Individuals can use it via Dreamina (free daily credits, Japanese UI); developers via the BytePlus API from $0.035 per output image (USD billing).
Business problems it solves
About "Seedream"
What is Seedream
Seedream is a family of image generation and editing AI models developed by Seed, the AI research team of China's ByteDance. Beyond text-to-image generation, it is strong at reference-based editing and compositing and at rendering text inside images. As of August 2026, the latest generation is Seedream 5.0 (the lightweight Lite and the higher-end Pro).
Seedream is a model, not a standalone app, so there are multiple entry points. For individuals, the browser entry point is Dreamina, ByteDance's AI creative platform (international edition with a Japanese UI). Developers can call it through the international cloud BytePlus API from $0.035 per image (pay-as-you-go, billed in USD).
Where to use it — sorting out the four entry points
Researching Seedream surfaces several different names — Dreamina, Jimeng, BytePlus, Volcano Engine — which is confusing, so let's sort them out first.
| Entry point | Audience | Region | Details |
|---|---|---|---|
| Dreamina | General users | International (incl. Japan) | Use Seedream 5.0 Lite / 5.0 Pro in the browser. Free daily credits |
| Jimeng AI | General users | China | The Chinese edition of Dreamina |
| BytePlus ModelArk | Developers | International | Pay-as-you-go API. Seedream 5.0 Pro / 5.0 Lite / 4.5 / 4.0, plus editing-focused SeedEdit 3.0 |
| Volcano Engine | Developers | China | API via ByteDance's China-facing cloud |
From Japan, Dreamina and BytePlus are the two realistic options. The rest of this page focuses on those two.
How to use
-
Create a Dreamina account
Sign up at Dreamina (the same ByteDance account system as CapCut). The UI can be switched to Japanese. Free daily credits are granted every day, so you can try it with no upfront payment.
-
Choose Seedream as the model in image generation
In the image generation screen, select Seedream 5.0 Lite / 5.0 Pro or similar as the model. Dreamina is a multi-model platform, so models other than Seedream are also listed.
-
Enter a prompt and reference images
Describe the image you want in text. You can also upload your own images and edit them — for example, changing the background or materials while keeping the subject.
-
Generate, then refine with local edits
With Seedream 5.0 Pro you can make local edits by pointing or sketching where to change, swap colors and materials, and separate elements into layers for adjustment. Download the result when done.
-
Move to the API for production systems
Create an account at BytePlus, issue a ModelArk API key, and call the image generation API. SDKs (Python / Go / Java, etc.) are provided, and new users get a free quota.
Key features
01Generate
- Text to image — Generate images up to 2K resolution from prompts. Strong at reproducing photographic lighting and textures
- Multilingual text rendering — Render poster and banner copy directly inside images. Officially described as supporting input and rendering for "over ten commonly used languages worldwide"
- Infographic generation — Turn data and long-form content into laid-out, diagrammatic images (5.0 Pro)
02Edit
- Instruction-based image editing — Edit uploaded images with instructions like "remove X" or "change Y"
- Interactive precision editing — Specify the target with point selection, lasso selection, or sketching, and repaint only that area (5.0 Pro)
- Layer separation — Decompose subject, background, and text into independent layers output as individually editable assets (5.0 Pro)
03Combine
- Multi-image compositing — Pass multiple reference images and perform composites like product swaps and element transfers in a single instruction
- Sequential generation — Generate a series of images while keeping the same world and characters (API)
- Built-in online search — On Dreamina, generation can incorporate web search context (5.0 Pro)
Pricing
Pricing differs by entry point. All figures below are based on official listings as of August 19, 2026.
Dreamina (individuals)
Free daily credits are granted every day, and the Seedream 5.0 models can be tried with these credits. Paid subscriptions exist, but plan names and prices are not listed on public pages and are shown only after logging in (not disclosed; check after login).
BytePlus ModelArk API (developers, pay-as-you-go in USD)
| Model | Per output image | Input (reference images) |
|---|---|---|
| Seedream 5.0 Pro | 1.5K or lower (≤2.61MP) $0.045 / above that $0.09 | 1st free, from 2nd $0.003/image |
| Seedream 5.0 Pro (layer decomposition) | 1.5K or lower $0.0225 / above that $0.045 | Same as above |
| Seedream 5.0 Lite | $0.035 | Free |
| Seedream 4.5 | $0.04 | Free |
| Seedream 4.0 | $0.03 | Free |
| SeedEdit 3.0 (image editing) | $0.03 | Free |
Images that fail to output due to moderation or similar reasons are not billed. Sequential generation is billed by the number of images actually generated. Tax treatment depends on your contracting region, so check billing terms in the BytePlus console. There is no JPY price list at any entry point. Payments with a Japanese card are converted at the exchange rate at the time of settlement.
Free plan and trial
- Dreamina — Each account receives free daily credits every day. You can try image generation with no upfront payment, but producing at volume assumes a paid plan. Whether free outputs carry a watermark and whether commercial use is allowed are not stated on public pages, so check at export time.
- BytePlus — A free trial of 200 images for Seedream models is offered to new users (official site as of August 19, 2026).
Japanese language support
- UI — Dreamina officially provides a Japanese UI (dreamina.capcut.com/ja-jp). The BytePlus console and documentation are in English.
- Input — Prompts are multilingual, and instructions can be given in Japanese. For text rendered inside images, the official description says "over ten languages," but it does not state whether Japanese is on the list. Before using Japanese lettering in production, verify accuracy on actual outputs.
- Support — No official Japanese-language support channel could be confirmed.
Reputation and track record
On third-party public benchmarks, Seedream 5.0 Pro sits at roughly 1280 ELO, ranked 8th overall in the Artificial Analysis Text to Image arena and 6th in the Image Editing category (as of August 19, 2026 — Artificial Analysis). The same site lists its generation cost at $90 per 1,000 images (high resolution), which is on the low-cost side among top-ranked models.
ByteDance Seed's official blog frames 5.0 Pro as a model that "doesn't just generate — it understands design." At the same time, the company itself notes "room to improve in finer-grained text rendering and pixel-level editing consistency."
For the international edition, no customer names or user counts have been published.
How it differs from other tools
When comparing image generation models, editing/compositing strength and pricing structure are the deciding factors.
| Aspect | Seedream 5.0 | Nano Banana Pro | FLUX | Midjourney |
|---|---|---|---|---|
| Provider | ByteDance (China) | Google (US) | Black Forest Labs (Germany) | Midjourney (US) |
| Consumer entry point | Dreamina (Japanese UI) | Gemini app | Playground etc. | Dedicated web/Discord |
| Editing | Precision editing, layer separation | Strong conversational editing | Editing via Kontext | Generation-focused |
| Approx. API price | From $0.035/image (official) | Pay-as-you-go API | 1 credit = $0.01 | Limited API |
| Pricing entry | Free daily credits | Free tier | Pay-as-you-go | Subscription |
The main comparisons are Google's Nano Banana, strong at conversational editing and text rendering; FLUX, known for photorealism and open weights; Ideogram, specialized in in-image text; and Midjourney, with its distinctive artistic style. As Chinese-affiliated services, Kling and Hailuo share the same points to weigh (billing currency, data handling).
FAQ
Q. Can generated images be used commercially?
Under Dreamina's terms of service, as long as you comply with the terms, ownership of uploaded assets and generated outputs belongs to the user. However, the same terms also state the platform is generally provided "for private, non-commercial use," and grant ByteDance a broad license to use your assets. For business use or client deliverables, review the terms in force at contract time and consider going through the BytePlus API, which is designed for commercial use.
Q. Are outputs watermarked?
The API has a watermark parameter that controls whether a watermark is added to output images; the official documentation's sample code calls the API with it disabled. Whether Dreamina's free outputs are watermarked, and whether an invisible watermark is embedded, are not stated on official public pages (as of August 19, 2026).
Q. Are uploaded images used for training? Where is data stored?
Dreamina's privacy policy states that collected information is used "to train and improve our machine learning models and algorithms." It also states that face and body features extracted from images and videos are deleted as soon as the service is provided and are not shared with third parties. Data is stored on servers in the United States, Singapore, and Malaysia. For sensitive material, review the terms and policy in force at contract time.
Q. Can I pay in Japanese yen?
There is no JPY price list. The BytePlus API is pay-as-you-go in USD; payments with a Japanese card are converted at the settlement-time exchange rate. The display currency of Dreamina's paid plans could not be confirmed on public pages.
Q. Can I cancel or get a refund?
Dreamina's terms of service and public pages contain no confirmed details on subscription cancellation procedures or refund conditions (as of August 19, 2026). Before subscribing to a paid plan, check the renewal and cancellation terms on the checkout screen.
Q. What's the difference between Seedream and Seedance?
Both are ByteDance Seed model families: Seedream is for images, Seedance is for video.
Provider
Seedream is developed by "ByteDance Seed," the AI research organization of ByteDance, headquartered in Beijing, China. ByteDance operates TikTok, Douyin, and CapCut and is one of the world's largest tech companies, privately held (not listed). Its international services are run by Singapore entities: Dreamina's operator and data controller is Bytedance Pte. Ltd. (1 Raffles Quay, Singapore), and the API is provided by BytePlus Pte Ltd. The Seed family also includes related models such as Seedance (video generation) and SeedEdit (image editing).
The information on this page is based on what is published on official sites as of August 2026. Features and pricing may change, so please check the official site for the latest information.




