0
# Chat Completions
1

2
Conversational AI interface supporting chat templates, tool calling, and multi-turn conversations with proper message formatting and context management. Provides OpenAI-compatible chat completion functionality.
3

4
## Capabilities
5

6
### Chat Interface
7

8
Generate responses in conversational format with support for system messages, user messages, assistant messages, and advanced features like tool calling.
9

10
```python { .api }
11
def chat(
12
    self,
13
    messages: List[ChatCompletionMessageParam],
14
    chat_template: Optional[str] = None,
15
    chat_template_content_format: ChatTemplateContentFormatOption = "auto",
16
    add_generation_prompt: bool = True,
17
    continue_final_message: bool = False,
18
    tools: Optional[List[ChatCompletionToolParam]] = None,
19
    documents: Optional[List[ChatCompletionDocumentParam]] = None,
20
    mm_processor_kwargs: Optional[Dict[str, Any]] = None,
21
    **kwargs: Any,
22
) -> List[ChatCompletionOutput]:
23
    """
24
    Generate chat completions from conversation messages.
25

26
    Parameters:
27
    - messages: List of conversation messages with roles and content
28
    - chat_template: Custom chat template for message formatting
29
    - chat_template_content_format: Content format handling ("auto", "string", "openai")
30
    - add_generation_prompt: Whether to add generation prompt
31
    - continue_final_message: Continue from the last assistant message
32
    - tools: Available tools for function calling
33
    - documents: Context documents for retrieval
34
    - mm_processor_kwargs: Multimodal processing arguments
35

36
    Returns:
37
    List of ChatCompletionOutput objects with generated responses
38
    """
39
```
40

41
## Usage Examples
42

43
### Basic Chat Conversation
44

45
```python
46
from vllm import LLM
47

48
llm = LLM(model="microsoft/DialoGPT-medium")
49

50
messages = [
51
    {"role": "system", "content": "You are a helpful assistant."},
52
    {"role": "user", "content": "What is the capital of France?"},
53
]
54

55
response = llm.chat(messages)
56
print(response[0].message.content)
57
```
58

59
## Types
60

61
```python { .api }
62
class ChatCompletionMessageParam:
63
    role: str  # "system", "user", "assistant"
64
    content: str  # Message content
65

66
class ChatCompletionOutput:
67
    message: ChatMessage  # Generated response
68
    finish_reason: Optional[str]  # Completion reason
69
```