Overview Generate images from text prompts using any of LiteLLM’s supported image generation providers. Returns responses in OpenAI format.
Function Signature def image_generation ( prompt : str , model : Optional[ str ] = None , n : Optional[ int ] = None , quality : Optional[Union[ str , ImageGenerationRequestQuality]] = None , response_format : Optional[ str ] = None , size : Optional[ str ] = None , style : Optional[ str ] = None , user : Optional[ str ] = None , timeout : float = 600 , api_key : Optional[ str ] = None , api_base : Optional[ str ] = None , api_version : Optional[ str ] = None , custom_llm_provider : Optional[ str ] = None , ** kwargs ) -> ImageResponse
Parameters Required Parameters Text description of the desired image(s). prompt = "A serene landscape with mountains and a lake at sunset"
Optional Parameters The model to use for image generation. Examples:
dall-e-3 (OpenAI)dall-e-2 (OpenAI)stability-ai/stable-diffusion-xl-base-1.0 (Bedrock)imagegeneration@006 (Vertex AI) Number of images to generate. Note: dall-e-3 only supports n=1
Size of generated images. For DALL-E 3:
"1024x1024" (default)"1792x1024""1024x1792" For DALL-E 2:
"256x256""512x512""1024x1024" Quality of the image (DALL-E 3 only). Options:
"standard": Standard quality"hd": Higher quality, more detailed Style of generated images (DALL-E 3 only). Options:
"vivid": Hyper-real and dramatic images"natural": More natural, less hyper-real Format of the returned image. Options:
"url": Returns a URL to the image"b64_json": Returns base64 encoded JSON Unique identifier for your end-user.
Request timeout in seconds (default 10 minutes).
API Configuration API key for the provider.
Base URL for the API endpoint.
Override provider detection.
Response ImageResponse Unix timestamp of when the image was created.
List of generated image objects. URL of the generated image (if response_format=“url”).
Base64 encoded image (if response_format=“b64_json”).
The revised prompt used by DALL-E 3 (if applicable).
Usage Examples Basic Image Generation import litellm response = litellm.image_generation( prompt = "A cute baby sea otter" , model = "dall-e-3" ) print (response.data[ 0 ].url)
High Quality Image import litellm response = litellm.image_generation( prompt = "A futuristic cityscape at night" , model = "dall-e-3" , quality = "hd" , size = "1792x1024" , style = "vivid" ) print (response.data[ 0 ].url) print (response.data[ 0 ].revised_prompt) # See how DALL-E revised your prompt
Multiple Images (DALL-E 2) import litellm response = litellm.image_generation( prompt = "A white siamese cat" , model = "dall-e-2" , n = 4 , size = "512x512" ) for i, image in enumerate (response.data): print ( f "Image { i + 1 } : { image.url } " )
Async Image Generation import litellm import asyncio async def main (): response = await litellm.aimage_generation( prompt = "A serene mountain landscape" , model = "dall-e-3" ) print (response.data[ 0 ].url) asyncio.run(main())
Base64 Response import litellm import base64 from PIL import Image from io import BytesIO response = litellm.image_generation( prompt = "A colorful abstract painting" , model = "dall-e-2" , response_format = "b64_json" ) # Decode and save the image image_data = base64.b64decode(response.data[ 0 ].b64_json) image = Image.open(BytesIO(image_data)) image.save( "generated_image.png" )
Provider Examples OpenAI DALL-E import litellm # DALL-E 3 response = litellm.image_generation( prompt = "A photorealistic cat" , model = "dall-e-3" , size = "1024x1024" , quality = "hd" ) # DALL-E 2 response = litellm.image_generation( prompt = "A photorealistic cat" , model = "dall-e-2" , n = 2 , size = "512x512" )
Azure OpenAI import litellm response = litellm.image_generation( prompt = "A beautiful sunset" , model = "azure/dall-e-3" , api_key = "your-azure-key" , api_base = "https://your-endpoint.openai.azure.com/" , api_version = "2024-02-01" )
AWS Bedrock import litellm response = litellm.image_generation( prompt = "A majestic eagle in flight" , model = "bedrock/stability.stable-diffusion-xl-v1" , # Bedrock-specific params width = 1024 , height = 1024 , cfg_scale = 7.0 , steps = 50 )
Vertex AI import litellm response = litellm.image_generation( prompt = "A cyberpunk city" , model = "vertex_ai/imagegeneration@006" , # Vertex-specific params number_of_images = 1 , aspect_ratio = "1:1" )
Replicate import litellm response = litellm.image_generation( prompt = "A fantasy castle" , model = "replicate/stability-ai/sdxl" , custom_llm_provider = "replicate" )
Saving Images Save from URL import litellm import requests from PIL import Image from io import BytesIO response = litellm.image_generation( prompt = "A peaceful garden" , model = "dall-e-3" ) # Download and save image_url = response.data[ 0 ].url image_response = requests.get(image_url) image = Image.open(BytesIO(image_response.content)) image.save( "generated_image.png" )
Save from Base64 import litellm import base64 response = litellm.image_generation( prompt = "An abstract artwork" , model = "dall-e-2" , response_format = "b64_json" ) # Save directly with open ( "image.png" , "wb" ) as f: f.write(base64.b64decode(response.data[ 0 ].b64_json))
Error Handling import litellm from litellm import ( BadRequestError, AuthenticationError, RateLimitError, ContentPolicyViolationError ) try : response = litellm.image_generation( prompt = "A beautiful landscape" , model = "dall-e-3" ) except ContentPolicyViolationError as e: print ( f "Content policy violation: { e } " ) except AuthenticationError as e: print ( f "Authentication failed: { e } " ) except RateLimitError as e: print ( f "Rate limit exceeded: { e } " ) except BadRequestError as e: print ( f "Bad request: { e } " ) except Exception as e: print ( f "An error occurred: { e } " )
Supported Providers LiteLLM supports image generation from:
OpenAI : DALL-E 2, DALL-E 3Azure OpenAI : DALL-E 2, DALL-E 3AWS Bedrock : Stable Diffusion, Titan Image GeneratorGoogle Vertex AI : Imagen 2, Imagen 3Replicate : Stable Diffusion XL, Flux, and moreTogether AI : Various Stable Diffusion models
See Image Generation Providers for the complete list.