Documentation

An outcome-focused guide to what you can do with Howlet Studio, and how easily.

LLM Node

The LLM Node connects to language model providers and generates text completions based on input prompts.

Supported Providers

OpenAI
GPT-4o, o1
Anthropic
Claude Opus 4.7, Sonnet, Haiku
Local Models
Ollama, vLLM, TGI
Azure OpenAI
Azure-hosted OpenAI models
AWS Bedrock
Titan, Claude and 30+ model profiles
Hugging Face
Any compatible model from HF Hub

Configuration Parameters

ParameterDescriptionDefault
modelSelect the model to use-
api_keyYour provider API key (stored encrypted)-
temperature0.0 (deterministic) to 2.0 (creative)0.7
max_tokensMaximum output length1024
system_messageSystem prompt for role/behavior instructions-
streamingReal-time token streamingfalse

Inputs & Outputs

Input

Text (prompt string) or Message (structured chat message)

Output

Text (generated completion) or Message (structured response with metadata)

The LLM Node can be chained with Prompt Template nodes for dynamic prompt construction.