Skip to main content

Using Image Generation in elvex

Understand how to setup and use image generation in elvex

Image generation on elvex is only available to customers who have configured an image generation provider.

Step 1: Set up an Image Generation Provider

Before using image generation, an Owner or Admin must configure an image generation model:

  1. Go to Admin Settings > Agents to access your image generation configuration

  2. Under Image generation provider, select a provider from the dropdown

  3. Press the Save button to apply your selection

Note: elvex supports multiple image generation providers, including OpenAI, Azure AI Foundry, and Google Gemini. Your company admin selects one provider for the entire company, and every agent with image generation enabled will use models from that provider.

Once an image generation provider is configured, users can generate images in any thread — simply describe the image you want in your message.

To give a specific agent the ability to generate images, go to the agent's settings, expand the Actions section, and enable the Image Generation action.

Supported Models

Different image providers offer different models, and each model has its own strengths — some are faster and better suited for everyday image creation, while others provide finer control for detailed edits or specific formatting needs like transparent backgrounds. Knowing which models are available under your provider helps you pick the right one for the task, or understand why an agent's output looks the way it does.

You might need to know which model is being used when:

  • You're troubleshooting why a generated image doesn't have a transparent background or didn't apply an edit as precisely as expected

  • You want to generate multiple images or edit several images at once and need to know the model's limits

  • You're deciding between providers for your organization and want to compare capabilities before configuring one

OpenAI Models

If your organization uses OpenAI as its image generation provider, the following models are available:

Model

Create images

Edit images

Transparent background

Edit more than one image

GPT Image 2.5 Flare

Yes (default for creating)

Yes

Yes

Yes, up to 16

GPT Image 2.5 Sunburst

Yes

Yes (default for editing)

Yes

Yes, up to 16

GPT Image 2

Yes

Yes

No

Yes, up to 16

GPT Image 1.5

Yes

Yes

Yes

No

GPT Image 1

Yes

Yes

Yes

No

  • Flare is fast and delivers high quality for most everyday tasks. elvex uses Flare by default when creating new images.

  • Sunburst produces the most detail and the most precise edits. elvex uses Sunburst by default when editing existing images.

  • Available image sizes: square (1024×1024), landscape (1536×1024), or portrait (1024×1536).

Azure AI Foundry

Azure AI Foundry works differently from OpenAI and Google Gemini: Azure always uses the model your admin configured on the Azure deployment itself. You cannot select a different model from within elvex — to change the model, your admin must update the deployment configuration in Azure.

Google Gemini Models

If your organization uses Google Gemini as its image generation provider, the following models are available:

Model

Create images

Edit images

Aspect ratio control

Input images for edits

Gemini 3.1 Flash Image (preview)

Yes (default)

Yes (default)

Yes

Up to 10 objects or 4 people

Gemini 3 Pro Image (preview)

Yes

Yes

Yes

Up to 14 (5 at high fidelity)

Gemini 2.5 Flash Image

Yes

Yes

No

Best results with 3 or fewer

  • Available aspect ratios: 1:1, 3:2, 2:3, 4:3, 3:4, 16:9, and 9:16.

Selecting a Model

elvex uses the default model for your configured provider automatically, so you don't need to do anything to get started. If you want to use a different available model for a specific request, just name it in your message. For example: "Use GPT Image 2.5 Sunburst to make a detailed poster."

What You Can Edit

When asking elvex to edit an image, you can reference:

  • Images that elvex created or edited earlier in the conversation

  • Images that you uploaded to the conversation

  • Images in artifacts in the conversation

Crafting Effective Prompts

To get the best results from image generation:

  • Be specific in your descriptions

  • Include details about style, lighting, and composition

  • Use adjectives to convey mood and atmosphere

Example: "Generate an image of a futuristic cityscape at sunset, with flying cars and neon signs, in a cyberpunk style"

Iterate on Generated Images

If the first result isn't quite what you wanted:

  • Analyze what aspects of the image you'd like to change

  • Modify your prompt to address these aspects

  • Try generating the image with a different model

  • Generate a new image with the updated prompt

Image Generation Parameters

elvex supports multiple image generation providers and models, each with its own capabilities. If you want to change a parameter for an agent, just use natural language to describe the parameter. For example: "Generate a landscape image of a highway interchange" or "Create an image of a cat with a transparent background"

Troubleshooting

Common Issues and Solutions

OpenAI Error: "Need to verify your organization settings"

Related Articles
Did this answer your question?