Models
CogView-4
Zhipu AI's image generation model with strong Chinese/English text rendering and multiple resolution support.
Overview
CogView-4 is Zhipu AI's image generation model, designed for precise and personalized AI visual expression. It features strong capabilities in complex semantic alignment and instruction following, supports both Chinese and English input of any length, and can generate images at various resolutions.
Key Specifications
| Parameter | Value |
|---|---|
| Input | Text prompt |
| Output | Generated image |
| Billing | Per request |
| Resolutions | 1024x1024, 768x1344, 864x1152, 1344x768, 1152x864, 1440x720, 720x1440 |
Capabilities
- Text-to-Image — Generate images from text descriptions
- Chinese/English Prompts — Native Chinese prompt support
- Multiple Resolutions — Various aspect ratios supported
- Quality Options — Standard (~5-10s) and HD (~20s)
Pricing
| Type | Price |
|---|---|
| Per Image | $0.008 / request |
Usage Example
response = client.images.generate(
model="cogview-4",
prompt="A cat sitting on a windowsill at sunset",
size="1024x1024",
)Best Practices
- Use standard quality for fast iterations and prototyping
- Use HD quality for final production outputs
- Keep prompts descriptive and specific for best results
- Chinese prompts work exceptionally well — a key differentiator