OpenAI Images

Generate images with DALL-E and GPT-Image

Generate images with OpenAI’s image models — DALL-E 2, DALL-E 3, and GPT-Image.

Generate

from praisonaiagents import ImageAgent

agent = ImageAgent(llm="openai/dall-e-3")
result = agent.generate("A sunset over mountains")
print(result.data[0].url)

GPT-Image

gpt-image-1 always returns the image as base64.

from praisonaiagents import ImageAgent

agent = ImageAgent(llm="openai/gpt-image-1")
result = agent.generate("A sunset over mountains")
print(result.data[0].b64_json[:40], "...")

With Options

from praisonaiagents import ImageAgent

agent = ImageAgent(llm="openai/dall-e-3")
result = agent.generate(
    "A futuristic city",
    size="1792x1024",
    quality="hd",
    style="vivid"
)
print(result.data[0].url)

Response Format

Default is url. Set b64_json to receive the image inline as base64.

from praisonaiagents import ImageAgent

agent = ImageAgent(llm="openai/dall-e-3", response_format="b64_json")
result = agent.generate("A red circle icon")
print(result.data[0].b64_json[:40], "...")

The default response_format="url" is not sent to OpenAI, so models that reject the parameter, such as gpt-image-1, work without extra configuration.

Models

Model	Generate	Edit	Variation
openai/gpt-image-1	✅	❌	❌
openai/dall-e-3	✅	❌	❌
openai/dall-e-2	✅	✅	✅