ByteDance / seedream-v4
seedream-v4
Access the Seedream 4.0 API for fast high resolution image generation and editing. Simple integration, clear pricing, and support for text to image, multi image input, and advanced creative workflows.
Seedream 4.0 is an advanced image generation system that can create images from text, edit existing images, and combine multiple images in one place. It uses an efficient transformer model with a powerful VAE that reduces the number of tokens it needs. This lets the system train more efficiently and generate high resolution images from 1K to 4K much faster than older models.
The model was trained on billions of text image pairs across many topics, styles, and real world scenarios. This large and diverse training data helps Seedream 4.0 understand a wide range of subjects and produce stable, high quality results. A tuned vision language model is also used so that text to image generation and image editing can be trained together for better overall performance.
Seedream 4.0 includes several speed improvements such as distillation, quantization, and speculative decoding. These help the model generate a 2K image in about 1.8 seconds. It performs strongly on image creation and multimodal editing tasks, handling complex prompts, detailed edits, multiple reference images, and multi image outputs. This makes Seedream 4.0 a powerful creative tool for both everyday use and professional work.
Example prompts
Great starting points for seedream-v4.
API quick start
Run seedream-v4 with a single API call.
{
"prompt": "A serene desert landscape at dusk, a lone…",
"size": "2048x2048",
"maxImages": 1,
"watermark": "false"
}