채팅 컴플리션, Responses API에서의 이미지 생성

채팅 컴플리션, Responses API에서의 이미지 생성 (Image Generation in Chat Completions, Responses API)

이 가이드는 chat/completions 사용 시 이미지를 생성하는 방법을 다뤄요. 참고: Responses API에서 원한다면 여기에서 Feature Request를 제출하세요.

LiteLLM v1.76.1+ 필요

지원 프로바이더:

• Google AI Studio (gemini)
• Vertex AI (vertex_ai/)

LiteLLM은 채팅 컴플리션 중 이미지 생성을 지원하는 모델의 assistant 메시지에서 images 응답을 표준화합니다.

"message": {
    ...
    "content": "Here's the image you requested:",
    "images": [
        {
            "image_url": {
                "url": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAA...",
                "detail": "auto"
            },
            "index": 0,
            "type": "image_url"
        }
    ]
}

빠른 시작

• SDK
• PROXY

from litellm import completion
import os 

os.environ["GEMINI_API_KEY"] = "your-api-key"

response = completion(
    model="gemini/gemini-2.5-flash-image-preview",
    messages=[
        {"role": "user", "content": "Generate an image of a banana wearing a costume that says LiteLLM"}
    ],
)

print(response.choices[0].message.content)  # Text response
print(response.choices[0].message.images)   # List of image objects

1. config.yaml 설정

model_list:
  - model_name: gemini-image-gen
    litellm_params:
      model: gemini/gemini-2.5-flash-image-preview
      api_key: os.environ/GEMINI_API_KEY

2. proxy server 실행

litellm --config config.yaml

# RUNNING on http://0.0.0.0:4000

3. 테스트!

curl http://0.0.0.0:4000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -d '{
    "model": "gemini-image-gen",
    "messages": [
      {
        "role": "user",
        "content": "Generate an image of a banana wearing a costume that says LiteLLM"
      }
    ]
  }'

기대 응답:

{
    "id": "chatcmpl-3b66124d79a708e10c603496b363574c",
    "choices": [
        {
            "finish_reason": "stop",
            "index": 0,
            "message": {
                "content": "Here's the image you requested:",
                "role": "assistant",
                "images": [
                    {
                        "image_url": {
                            "url": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAA...",
                            "detail": "auto"
                        },
                        "index": 0,
                        "type": "image_url"
                    }
                ]
            }
        }
    ],
    "created": 1723323084,
    "model": "gemini/gemini-2.5-flash-image-preview",
    "object": "chat.completion",
    "usage": {
        "completion_tokens": 12,
        "prompt_tokens": 16,
        "total_tokens": 28
    }
}

출처: 문서

본문

스트리밍 지원

• SDK
• PROXY

from litellm import completion
import os 

os.environ["GEMINI_API_KEY"] = "your-api-key"

response = completion(
    model="gemini/gemini-2.5-flash-image-preview",
    messages=[
        {"role": "user", "content": "Generate an image of a banana wearing a costume that says LiteLLM"}
    ],
    stream=True,
)

for chunk in response:
    if hasattr(chunk.choices[0].delta, "images") and chunk.choices[0].delta.images is not None:
        print("Generated image:", chunk.choices[0].delta.images[0]["image_url"]["url"])
        break

curl http://0.0.0.0:4000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -d '{
    "model": "gemini-image-gen",
    "messages": [
      {
        "role": "user",
        "content": "Generate an image of a banana wearing a costume that says LiteLLM"
      }
    ],
    "stream": true
  }'

기대 스트리밍 응답:

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1723323084,"model":"gemini/gemini-2.5-flash-image-preview","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]}

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1723323084,"model":"gemini/gemini-2.5-flash-image-preview","choices":[{"index":0,"delta":{"content":"Here's the image you requested:"},"finish_reason":null}]}

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1723323084,"model":"gemini/gemini-2.5-flash-image-preview","choices":[{"index":0,"delta":{"images":[{"image_url":{"url":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAA...","detail":"auto"},"index":0,"type":"image_url"}]},"finish_reason":null}]}

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1723323084,"model":"gemini/gemini-2.5-flash-image-preview","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Async 지원

from litellm import acompletion
import asyncio
import os 

os.environ["GEMINI_API_KEY"] = "your-api-key"

async def generate_image():
    response = await acompletion(
        model="gemini/gemini-2.5-flash-image-preview",
        messages=[
            {"role": "user", "content": "Generate an image of a banana wearing a costume that says LiteLLM"}
        ],
    )
    
    print(response.choices[0].message.content)  # Text response
    print(response.choices[0].message.images)   # List of image objects

    return response

# Run the async function
asyncio.run(generate_image())

지원 모델

스펙

응답의 images 필드는 다음 구조를 따릅니다:

"images": [
    {
        "image_url": {
            "url": "data:image/png;base64,<base64_encoded_image>",
            "detail": "auto"
        },
        "index": 0,
        "type": "image_url"
    }
]

• images - List[ImageURLListItem]: 생성된 이미지 배열
• image_url - ImageURLObject: 이미지 데이터 컨테이너
  • url - str: data URI 형식의 base64 인코딩 이미지 데이터
  • detail - str: 이미지 상세 수준 (생성된 이미지에서는 항상 "auto")
  • index - int: 응답에서 이미지의 인덱스
  • type - str: 타입 식별자 (항상 "image_url")

이미지는 base64로 인코딩된 data URI로 반환되며, HTML <img> 태그에 직접 사용하거나 파일로 저장할 수 있어요.

더 알아보기 (Learn more)

• Chat Completions
• Image Generation