GPT Image 1 is our new state-of-the-art image generation model. It is a natively multimodal language model that accepts both text and image inputs, and produces image outputs.
How to Use GPT-Image-1 for Image Generation in LibreChat
LibreChat now supports image creation using GPT-Image-1. Follow the steps below to set up and start generating images:
Recommended Platform: Use JuheNext AI model aggregation platform: https://www.juhenext.com
LibreChat Access: Login to LibreChat at: https://librechat.aijuhe.top/login
Assistant Configuration
-
Register and Log In
- Register and log in to the LibreChat application deployed via JuheNext.
-
Set OpenAI API Key
- You must configure your OpenAI API key at least once.
- After setting the API key, select any AI model under the OpenAI group and start a conversation. If you get responses, your setup is successful.
-
Set Up Agent
- In the right sidebar, find “Agent Builder” and name the agent “GPT-IMAG-1.”
- Select one of the OpenAI models as the assistant Agent’s conversation model.
-
Add Image Tool
- Add the OpenAI image model as a tool, input your API key before adding.
- Save the changes.
Using Parameters for Image Generation
The drawing model supports both image generation and image editing, using OpenAI’s latest GPT-Image-1 model for superior instruction following, text rendering, detailed edits, and real-world knowledge.
1. Image Generation
- Generate a new image from a text prompt (no upload required).
- Supported parameters (you can describe in Chinese):
| Parameter | Description | Options |
|---|---|---|
| Prompt | Your description (required) | Text description |
| Size | Image dimensions | auto (default), 1024×1024 (square), 1536×1024 (landscape), 1024×1536 (portrait) |
| Quality | Image quality | auto (default), high, medium, low |
| Background | Background type | auto (default), transparent, opaque (PNG or WebP for transparent) |
Example Prompt: Draw an image of Superman fighting Iron Man, size 1536×1024, medium quality, transparent background, webp format.
(Note: Actual output size in LibreChat may be 1152×768 for 1536×1024 prompts.)
The AI will automatically extract the required parameters.
2. Image Editing
While GPT-Image-1 doesn’t truly modify an original image, it will reference the original elements and your requirements, recreating a new image that achieves the requested changes.
- Continue discussions to request edits; the model uses the current image ID for modifications or remixing.
- You can also upload an image as the base for new creations.
Cost Information
Compared to GPT-4o-image (reverse models), the API version GPT-Image-1 is more expensive, but offers greater speed and stability. For regular usage, GPT-4o-image is preferred for cost-savings.
| Model | Price (per image) |
|---|---|
| gpt-4o-image | ￥0.20 |
| gpt-image-1 | ￥1.00 – ￥2.50 |
Enjoy creating with GPT-Image-1 on LibreChat!
