Implementation:Infiniflow Ragflow DialogService Update LLM Settings

Knowledge Sources	RAGFlow
Domains	RAG, Conversational_AI
Last Updated	2026-02-12 06:00 GMT

Overview

Concrete tool for configuring LLM model and generation parameters on a chat application provided by RAGFlow DialogService.

Description

Uses DialogService.update_by_id to set llm_id and llm_setting fields. The LLM catalog (conf/llm_factories.json, 5646 lines) defines 66+ providers including OpenAI, DeepSeek, Moonshot, Tongyi-Qianwen, ZHIPU-AI, and many more.

Usage

Called via REST API to configure or change the LLM for a chat application.

Code Reference

Source Location

• Repository: ragflow
• File: api/apps/dialog_app.py (L34-144), conf/llm_factories.json

Signature

DialogService.update_by_id(pid: str, data: dict) -> int
# data = {
#     "llm_id": "gpt-4-turbo",
#     "llm_setting": {
#         "temperature": 0.1,
#         "top_p": 0.3,
#         "frequency_penalty": 0.7,
#         "presence_penalty": 0.4,
#         "max_tokens": 512
#     }
# }

Import

from api.db.services.dialog_service import DialogService

I/O Contract

Inputs

Name	Type	Required	Description
llm_id	str	No	Model ID (defaults to tenant default)
llm_setting.temperature	float	No	Default 0.1
llm_setting.top_p	float	No	Default 0.3
llm_setting.frequency_penalty	float	No	Default 0.7
llm_setting.presence_penalty	float	No	Default 0.4
llm_setting.max_tokens	int	No	Default 512

Outputs

Name	Type	Description
num	int	Rows updated

Usage Examples

from api.db.services.dialog_service import DialogService

DialogService.update_by_id("dialog-uuid-123", {
    "llm_id": "deepseek-chat",
    "llm_setting": {
        "temperature": 0.3,
        "max_tokens": 1024
    }
})

Related Pages

Implements Principle

• Principle:Infiniflow_Ragflow_LLM_Configuration