← Node reference

AI Chat

AI_LLM

Send a prompt to any registered LLM provider (OpenAI, Anthropic, Gemini, Groq, DeepSeek, Ollama) and store the response, or return structured JSON for downstream nodes.

Configuration

  • variableNametextoptional
  • modeltextoptional
  • fallbackModelslong textoptional
  • openaiCredentialIdcredentialoptional

    Accepts openai.apiKey. Stored encrypted; the value is never returned to the browser or written to a trace.

  • anthropicCredentialIdcredentialoptional

    Accepts anthropic.apiKey. Stored encrypted; the value is never returned to the browser or written to a trace.

  • geminiCredentialIdcredentialoptional

    Accepts gemini.apiKey. Stored encrypted; the value is never returned to the browser or written to a trace.

  • groqCredentialIdcredentialoptional

    Accepts groq.apiKey. Stored encrypted; the value is never returned to the browser or written to a trace.

  • systemPromptlong textoptional
  • userPromptlong textoptional
  • attachmentslong textoptional
  • temperaturenumberoptional
  • maxTokensnumberoptional
  • jsonModetrue / falseoptional
  • jsonSchemalong textoptional
  • cacheTtlSecondsnumberoptional

Ports

Inputs

  • main In

Outputs

  • main Out

Execution

Type version
1
Retry
no retry by default
Timeout
engine default
Response cache
Supported — set cacheTtlSeconds to reuse an identical call within the window. A cached run records zero tokens and zero cost, because nothing was purchased.
Credentials
openai.apiKey (optional), anthropic.apiKey (optional), gemini.apiKey (optional), groq.apiKey (optional)