Skip to content

Generate

LLMfy provides six generation methods. Use invoke for single-turn text/content input and chat for multi-turn message history.

Method Input Tool execution Streaming
invoke text / Content list No No
invoke_with_tools text / Content list Yes No
invoke_stream text / Content list No Yes
chat Message list No No
chat_with_tools Message list Yes No
chat_stream Message list No Yes

Invoke

invoke accepts plain text or a list of Content objects (supports images, documents, and video).

1
2
3
response = agent.invoke("Hello")

print(f"\n>> {response.result.content}\n")

Invoke With Tools

invoke_with_tools runs the tool-calling loop automatically — it calls registered tools until the model returns a final text response.

1
2
3
4
5
content = "What is the weather in London?"

response = agent.invoke_with_tools(content)

print(f"\n>> {response.result.content}\n")

Invoke Stream

Same as invoke but returns a generator that yields chunks as they arrive.

1
2
3
4
5
6
7
8
9
from llmfy import GenerationResponse

stream = agent.invoke_stream("Hello")
full_content = ""
for chunk in stream:
    if isinstance(chunk, GenerationResponse):
        if chunk.result.content:
            full_content += chunk.result.content
            print(chunk.result.content, end="", flush=True)

Chat

chat accepts a list of Message objects with explicit roles, enabling multi-turn conversations.

1
2
3
4
5
6
7
from llmfy import Message, Role

messages = [Message(role=Role.USER, content="Hello")]

response = agent.chat(messages)

print(f"\n>> {response.result.content}\n")

Chat With Tools

chat_with_tools runs the tool-calling loop automatically using message history.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
from llmfy import Message, Role

messages = [
    Message(
        role=Role.USER,
        content="What is the time and weather in London?",
    )
]

response = agent.chat_with_tools(messages)

print(f"\n>> {response.result.content}\n")

Chat Stream

Same as chat but returns a generator that yields chunks as they arrive.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from llmfy import Message, Role, GenerationResponse

messages = [Message(role=Role.USER, content="Hello")]

stream = agent.chat_stream(messages)
full_content = ""
for chunk in stream:
    if isinstance(chunk, GenerationResponse):
        if chunk.result.content:
            full_content += chunk.result.content
            print(chunk.result.content, end="", flush=True)