Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.xagent.run/llms.txt

Use this file to discover all available pages before exploring further.

​
Image Generation Models

Image generation models enable Xagent to create and modify images through text prompts.

​
Purpose

Enable Image Tools:
  • generate_image - Create images from text descriptions
  • edit_image - Modify existing images
  • Visual content creation and editing
How it works:
  • Accept text prompts describing desired image
  • Generate new images or modify existing ones
  • Return image files saved to workspace

​
When to Use

Required for:
  • Image generation tasks
  • Image editing and modification
  • Visual content creation
  • Design and marketing materials
Not required for:
  • Image understanding or analysis (uses Vision LLMs)
  • Image OCR or text extraction (uses Vision LLMs)
  • Chart/graph analysis (uses Vision LLMs)

​
Supported Providers

​
OpenAI & OpenAI-compatible

Models: gpt-image-1 and compatible models Setup:
  1. Get API key from OpenAI Platform
  2. For compatible services, provide base URL and API key
  3. Select image model
Abilities: Both “generate” and “edit” Best for:
  • High-quality image generation
  • Creative tasks
  • Marketing materials
  • Wide compatibility

​
DashScope (Alibaba Cloud)

Models: qwen-image Setup:
  1. Get API key from DashScope Console
  2. Configure API key
  3. Select model
Abilities: Primarily “generate” (can support “edit”) Best for:
  • Chinese language optimization
  • Cost-effective for Asian markets
  • Alibaba Cloud integration
Supported formats:
  • JPG, JPEG, PNG, BMP, TIFF, WEBP
  • Max file size: 10MB for editing

​
Gemini (Google)

Models: gemini-3-pro-preview-image Setup:
  1. Get API key from Google AI Studio
  2. Configure API key (supports GEMINI_API_KEY or GOOGLE_API_KEY)
  3. Select image model
Abilities: “generate” only Best for:
  • High-quality image generation
  • Google Cloud integration
  • Multiple resolution support
  • Various aspect ratios
Supported formats:
  • Aspect ratios: 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9
  • Image sizes: 1K (up to 1024px), 2K (up to 2048px), 4K (up to 4096px)
  • Output: PNG format (base64 encoded)
Features:
  • Text-to-image generation using Google’s Imagen
  • Automatic aspect ratio detection and selection
  • Resolution based on model capabilities (2K/4K)
  • Support for negative prompts
Gemini image models don’t support image editing. Use generate_image only.

​
Xinference

Models: Stable Diffusion variants (stable-diffusion-2-1, etc.) Setup:
  1. Deploy Xinference server
  2. Launch image generation model
  3. Configure base URL in Xagent
  4. Select model from Xinference
Abilities: Configurable (default: [“generate”], supports “edit”) Best for:
  • Self-hosted deployment
  • Data privacy
  • Cost control
  • Using open-source models
Features:
  • List available models from server
  • Support for various Stable Diffusion models
  • Image-to-image and inpainting capabilities

​
Configuration

​
Step 1: Add Image Provider

  1. Go to Models in the sidebar
  2. Click Add Model or Add Provider
  3. Select image generation provider
  4. Enter API credentials

​
Step 2: Configure Model

  1. Select the image model
  2. Configure parameters:
    • Image size (e.g., 10241024, 17921024)
    • Response format (url or b64_json)
    • Number of images (n)
  3. Test generation

​
Step 3: Set Abilities

Image models can have these abilities:
  • generate - Create new images from text
  • edit - Modify existing images
Not all image models support both generation and editing. Check provider documentation.

​
Usage Examples

​
Generating Images

User: "Create a promotional poster for a coffee shop"
Xagent: [Uses generate_image tool]
[Generates image based on description, saves to workspace]

​
Editing Images

User: "Change the color scheme to warm tones"
Xagent: [Uses edit_image tool]
[Modifies existing image, saves to workspace]

​
Multi-Image Editing

User: "Combine these two images with a sunset background"
Xagent: [Uses edit_image tool with multiple images]
[Edits multiple images into one result]

​
Troubleshooting

​
Generation Failed

Check:
  • API key is valid
  • Model supports generation ability
  • Prompt follows guidelines
  • Content policy violations

​
Poor Quality

Try:
  • Improving prompt specificity
  • Adding style instructions
  • Using negative prompts
  • Trying different model
  • Adjusting image size

​
Slow Generation

Optimize:
  • Reduce image size
  • Consider faster model
  • Check network connectivity

​
Edits Not Working

Verify:
  • Model supports edit ability
  • Edit instructions are clear
  • Original image is accessible
  • Image format is supported
  • File size within limits (DashScope: 10MB)

​
File Not Found

Check:
  • Image path is correct
  • File exists in workspace
  • Use workspace file browser to verify path
  • URL is accessible

​
Security & Privacy

Important considerations:
  • Generated images are saved to workspace
  • Check provider’s data retention policy
  • Be aware of content policies
  • Copyright and usage rights
Recommendations:
  • Review provider content policy
  • Ensure rights to generated content
  • Consider compliance requirements
  • Be mindful of copyright

​
Next Steps