BytePlus Models

BytePlus, ByteDance's cloud platform, offers cutting-edge AI models through AvalAI's unified API. BytePlus specializes in advanced image generation and editing capabilities, providing state-of-the-art models for creative and commercial applications.

Available Models

BytePlus models are accessible through AvalAI's standard endpoints, offering seamless integration with familiar OpenAI-compatible APIs while providing access to ByteDance's innovative AI technologies.

Image Generation Models

BytePlus offers the most advanced image generation models available, featuring both text-to-image and image-to-image capabilities with unprecedented quality and flexibility.

Seedream 5.0

Model ID: seedream-5-0-260128

Seedream 5.0 is BytePlus's most advanced image generation model, featuring Chain of Thought (CoT) reasoning for intelligent prompt optimization and built-in MJ-style aesthetic generation. This model represents a significant leap forward in AI-assisted creative generation.

Feature	Details
Model Type	Image Generation & Editing
Max Resolution	4K (4096x4096 pixels)
Supported Formats	JPEG, PNG
Input Methods	Text prompts, Image URLs, Base64
Batch Generation	Up to 15 images per request
Multi-Image Input	Up to 10 reference images
Streaming Support	Yes
CoT Reasoning	Built-in intelligent prompt optimization
Pricing	$0.035 per image ($35.00/1K images)

Key Capabilities

🧠 Chain of Thought Reasoning

• Intelligent Prompt Analysis: Analyzes prompts to understand artistic intent
• Automatic Enhancement: Optimizes prompts for better visual outcomes
• Style Recognition: Identifies and enhances desired artistic styles
• Composition Planning: Plans image composition before generation

🎨 MJ-Style Aesthetic Generation

• Built-in Aesthetics: Native support for Midjourney-style artistic generation
• Photorealistic Quality: Professional-grade photorealistic outputs
• Artistic Versatility: Support for various artistic styles and mediums
• Typography Excellence: Advanced text rendering with superior consistency

⚡ Technical Excellence

• High Resolution Support: Generate images up to 4K resolution
• Flexible Aspect Ratios: Support ratios from 1:16 to 16:1
• Real-time Streaming: Progressive image generation with live updates
• Smart Batch Processing: Automatic determination of optimal image count

Usage Examples

Text-to-Image Generation with CoT

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

response = client.images.generate(
    model="seedream-5-0-260128",
    prompt="A serene Japanese garden at sunset with cherry blossoms",
    size="4K",
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"Generated image: {response.data[0].url}")

Image Editing with CoT Reasoning

response = client.images.edit(
    model="seedream-5-0-260128",
    image="https://example.com/input-image.jpg",
    prompt="Transform into a professional portrait with studio lighting and subtle background blur",
    size="2K",
    response_format="url",
    extra_body={"sequential_image_generation": "disabled", "watermark": False},
)

print(f"Edited image: {response.data[0].url}")

---

Streaming Generation

import requests
import json


def stream_image_generation():
    url = "https://api.avalai.ir/v1/images/generations"
    headers = {
        "Authorization": "Bearer your-avalai-api-key",
        "Content-Type": "application/json",
    }

    data = {
        "model": "seedream-5-0-260128",
        "prompt": "A detailed digital artwork of a magical forest",
        "size": "2K",
        "stream": True,
        "sequential_image_generation": "auto",
        "sequential_image_generation_options": {"max_images": 3},
    }

    response = requests.post(url, headers=headers, json=data, stream=True)

    for line in response.iter_lines():
        if line:
            try:
                chunk = json.loads(line.decode("utf-8").replace("data: ", ""))
                if "data" in chunk:
                    for image in chunk["data"]:
                        if "url" in image:
                            print(f"Image ready: {image['url']}")
            except json.JSONDecodeError:
                continue


stream_image_generation()

Key Parameters

Core Parameters

Parameter	Type	Description	Default
model	string	Model identifier	Required
prompt	string	Text description (max 600 words)	Required
size	string	Resolution: "1K", "2K", "4K" or "WxH"	"2048x2048"
response_format	string	"url" or "b64_json"	"url"

BytePlus-Specific Parameters

Parameter	Type	Description	Default
sequential_image_generation	string	"auto" or "disabled"	"disabled"
sequential_image_generation_options	object	Batch generation settings	-
max_images	integer	Max images in batch (1-15)	15
stream	boolean	Enable streaming output	false
watermark	boolean	Add AI watermark	true
image	string/array	Reference image(s) for editing	-

Supported Resolutions

Aspect Ratio	Recommended Dimensions
1:1	2048x2048
4:3	2304x1728
3:4	1728x2304
16:9	2560x1440
9:16	1440x2560
3:2	2496x1664
2:3	1664x2496
21:9	3024x1296

Best Practices

Prompt Engineering

1. Be Descriptive: Use detailed prompts up to 600 English words
2. Style Specification: Include artistic style, mood, and technical details
3. Composition Details: Specify lighting, perspective, and visual elements
4. Quality Keywords: Use terms like "high quality", "detailed", "professional"

Technical Optimization

1. Resolution Planning: Choose appropriate resolution for your use case
2. Batch Generation: Use sequential generation for related image series
3. Streaming for Large Batches: Enable streaming for multiple image requests
4. Format Selection: Use URL format for web applications, Base64 for direct processing

Image Input Guidelines

1. Format Requirements: JPEG or PNG formats only
2. Size Limits: Maximum 10MB per image, 6000x6000 pixels
3. Aspect Ratios: Keep within 1:3 to 3:1 range
4. Multiple Images: Up to 10 reference images for blending

Use Cases

Creative Applications

• Digital Art Creation: High-quality artwork and illustrations
• Concept Art: Visual development for games, films, and products
• Style Transfer: Transform images into different artistic styles
• Creative Exploration: Generate variations and iterations of ideas

Commercial Applications

• Marketing Materials: Product visualizations and promotional graphics
• Brand Assets: Logo variations, brand imagery, and visual identity
• Content Creation: Social media graphics, blog illustrations
• E-commerce: Product mockups and lifestyle imagery

Technical Applications

• Prototyping: Quick visualization of design concepts
• Batch Processing: Generate series of related images efficiently
• Image Enhancement: Improve and transform existing imagery
• Custom Workflows: Integrate with existing creative pipelines

Integration with AvalAI

Seedream 5.0 integrates seamlessly with AvalAI's unified API ecosystem:

• OpenAI Compatibility: Use familiar OpenAI client libraries
• Standard Endpoints: v1/images/generations and v1/images/edits
• Unified Authentication: Single API key for all models
• Consistent Response Format: Standard JSON responses across all providers
• Error Handling: Unified error codes and messages

Limitations

1. Content Filtering: Images may be rejected by content safety filters
2. Generation Time: High-resolution images may take longer to generate
3. Batch Limits: Maximum 15 images per request (including reference images)
4. Language Support: Prompts work best in English
5. URL Expiration: Generated image URLs expire after 24 hours

Related Resources

• Image Generation Guide
• API Reference
• Seedream 5.0 Examples
• Pricing Information

---

BytePlus models are provided through ByteDance's cloud platform and integrated into AvalAI's unified API for seamless access and consistent developer experience.