Chat Model with Reasoning
AI-generatedSummary
Provides a chat language model for an AI Agent or AI Chain, using a selected OpenAI-compatible model endpoint. Optionally enables extended reasoning and can include returned reasoning text in the model output wrapped in
Inputs
- Model (required) — Model identifier selected from the endpoint’s model list; defaults to deepseek/deepseek-r1.
- Enable Reasoning — Enables reasoning parameters for supported models. When enabled, choose a reasoning effort (low, medium, or high) or set a reasoning token budget above 0; the token budget takes precedence.
- Include Reasoning in Output — When enabled with reasoning, requests reasoning text and prepends it to the response in
tags. When disabled, reasoning text is excluded from the response. - Options — Sets sampling temperature, final-response token limit, top-p, frequency and presence penalties, response format, request timeout, and retry count. Defaults include temperature 0.7, 4096 maximum output tokens, text format, 360000 ms timeout, and 2 retries. For JSON Object format, include “json” in the prompt.
Output shape
An AI language model connection for an AI Agent or AI Chain; it does not return a standalone item list.
Requires credentials with an API key and endpoint URL. The fetch wrapper also converts empty or invalid tool-call arguments to an empty JSON object.