Azure OpenAI ResponseInstall

Summary

Send a prompt or conversation messages to a specified Azure OpenAI model deployment to generate a stateful AI response optionally using tools and functions.

Inputs

• Model (required) — The deployment name of the Azure OpenAI model to use (e.g., gpt-4.1-mini, gpt-4o, o3-mini).
• Input (required) — The input for the model. Can be a plain string, a JSON string of messages (an array of role/content objects), or a message array to represent multi-turn conversation.
• Tools (Dynamic) — Optional dynamic tools/functions from external sources provided as JSON string or expression (e.g., {{ $json.tools }}).
• Additional Fields — Optional parameters including system instructions, previous response ID for continuing conversations, whether to store conversation history, temperature, top_p, maximum output tokens, reasoning effort level, background mode, streaming responses, included fields, and custom metadata.
• Tools — Optional built-in or custom tools/functions definitions to augment model capabilities, such as code interpreter, file search, image generation, or custom functions defined in JSON.
• Tool Options — Settings controlling how tools are used: whether to let the model decide, require, or disable tools, and whether to allow parallel tool calls.

Output shape

Multiple output streams: main 'Message' output with AI text and messages, a 'Tool' output for tool invocation results, and a 'Response' output with the complete API response object.

Outputs are distributed across three output channels: 'Message' contains conversational messages, 'Tool' contains tool call data if any tools were invoked, and 'Response' provides the full detailed response including metadata, usage, and errors if any. Supports handling stateful conversations by referencing previous responses and storing conversation history. Errors emit as outputs when 'Continue On Fail' is enabled.