Implementation:Langgenius Dify UpdateAppModelConfig

Knowledge Sources	Domains	Last Updated
Dify	LLM_Applications, Frontend, API	2026-02-12 00:00 GMT

Overview

Description

updateAppModelConfig is the frontend service function for persisting changes to an application's model configuration. It sends a POST request to a dynamically constructed URL (typically apps/{appId}/model-config) with the full model configuration body. The function accepts a generic Record<string, any> body to accommodate the varying shapes of model configuration across different application modes and LLM providers.

The body payload corresponds to the ModelConfig type, which encompasses the LLM provider, model ID, model mode, prompt configuration, completion parameters, feature toggles (speech-to-text, text-to-speech, more-like-this, suggested questions), dataset configurations, agent settings, and file upload settings.

Usage

Call updateAppModelConfig after modifying any aspect of the application's model or prompt configuration in the UI. This function persists the complete configuration state to the backend. It is typically invoked when the user saves changes in the application configuration panel.

Code Reference

Source Location

web/service/apps.ts, lines 156-158

Signature

export const updateAppModelConfig = (
  { url, body }: { url: string, body: Record<string, any> }
): Promise<UpdateAppModelConfigResponse> => {
  return post<UpdateAppModelConfigResponse>(url, { body })
}

Import

import { updateAppModelConfig } from '@/service/apps'

I/O Contract

Inputs

Parameter	Type	Required	Description
url	string	Yes	The API endpoint URL, typically apps/{appId}/model-config.
body	Record<string, any>	Yes	The full model configuration payload. Key properties include those described below.

Key properties within the body (ModelConfig shape):

Property	Type	Description
provider	string	The LLM provider identifier (e.g., 'openai', 'anthropic', 'azure_openai').
model_id	string	The specific model name (e.g., 'gpt-3.5-turbo', 'gpt-4').
mode	ModelModeType	The model interaction mode: 'chat' or 'completion'.
configs	PromptConfig	The prompt template and variable definitions: { prompt_template: string, prompt_variables: PromptVariable[] }.
completion_params	CompletionParams	Inference parameters: max_tokens, temperature (0-2), top_p (0-1), presence_penalty (-2 to 2), frequency_penalty (-2 to 2), stop (string[]).
chat_prompt_config	null	Chat-mode prompt messages array. Each entry has role and text.
completion_prompt_config	null	Completion-mode prompt with conversation history role configuration.
opening_statement	null	Initial greeting message for chat applications.
more_like_this	null	Toggle for generating alternative responses.
suggested_questions_after_answer	null	Toggle for suggesting follow-up questions.
speech_to_text	null	Toggle for speech-to-text input.
text_to_speech	null	Text-to-speech output configuration.
dataset_configs	DatasetConfigs	Retrieval strategy and connected dataset list.
agentConfig	AgentConfig	Agent mode settings including strategy (function_call or react), max iterations, and tool list.

Outputs

Field	Type	Description
(return)	Promise<UpdateAppModelConfigResponse>	Resolves to { result: string } indicating the outcome of the update operation (e.g., 'success').

Usage Examples

Updating model selection and completion parameters

import { updateAppModelConfig } from '@/service/apps'

const response = await updateAppModelConfig({
  url: `apps/${appId}/model-config`,
  body: {
    provider: 'openai',
    model_id: 'gpt-4',
    mode: 'chat',
    configs: {
      prompt_template: 'You are a helpful assistant specializing in {{domain}}.',
      prompt_variables: [
        { key: 'domain', name: 'Domain', type: 'string', required: true },
      ],
    },
    completion_params: {
      max_tokens: 2048,
      temperature: 0.7,
      top_p: 0.95,
      presence_penalty: 0.0,
      frequency_penalty: 0.0,
    },
    opening_statement: 'Hello! How can I help you today?',
    more_like_this: { enabled: false },
    suggested_questions_after_answer: { enabled: true },
    speech_to_text: { enabled: false },
    text_to_speech: { enabled: false },
  },
})

console.log(response.result) // 'success'

Related Pages

• Principle:Langgenius_Dify_Model_Configuration