stackit-ai-model-serving

n8n community nodes for STACKIT AI Model Serving (OpenAI-compatible Chat & Embeddings)

Package Information

Downloads: 0 weekly / 16 monthly
Latest Version: 1.1.0
Author: Andreas Klos

Available Nodes

STACKIT Chat Model icon
STACKIT Chat Model
Provides access to the STACKIT Chat Model API.
STACKIT Embeddings icon
STACKIT Embeddings
Provides access to the STACKIT Embeddings API.

Documentation

Project banner
Project banner

STACKIT AI Model Serving nodes for n8n

n8n community nodes that connect to STACKIT AI Model Serving’s OpenAI-compatible API. Includes:

  • STACKIT Chat Model — Chat-completions compatible LLM node.
  • STACKIT Embeddings — Text embeddings generation node.

n8n is a fair-code licensed workflow automation platform.

Installation

Follow the official guide to install community nodes: n8n Community Nodes Installation

Once installed, search for “STACKIT Chat Model” or “STACKIT Embeddings” in the node panel.

Credentials

These nodes use a single credential: “STACKIT AI Model Serving API”.

What you need:

How it works:

  • Set the API Key and (optionally) API URL in the credential "STACKIT AI Model Serving API".
  • All node requests use the credential’s API URL. If not set, the default endpoint above is used.
  • The Model picker fetches from ${API_URL}/models using your credential.

Notes:

  • These nodes are compatible with OpenAI-style models and endpoints provided by STACKIT

STACKIT Chat Model

Use this node to obtain a ChatCompletions-capable model as an AI Model output for n8n’s AI features or your own custom chains and agents. No LangChain dependency is required. The node accepts a variety of input shapes (including common LangChain-like prompt structures) and normalizes them for the API.

Key options:

  • Model: dynamically loaded from your STACKIT project (non-embedding models)
  • Temperature, Top P, Penalties, Max tokens, Retries, Timeout
  • Response format: text or JSON (when using JSON mode, ensure your prompts instruct the model to return valid JSON)

Typical usage:

  1. Add “STACKIT Chat Model” to the canvas
  2. Configure the node with your desired parameters (API-Key, Model and Options)
  3. Connect to AI nodes (e.g., AI Agent, AI Chain)

Behavior details:

  • Input normalization: The model function accepts strings, arrays of OpenAI-style messages, and objects like { input }, { messages }, { prompt }, as well as common LangChain prompt shapes (StringPromptValue/ChatPromptValue). These are normalized to OpenAI-compatible messages internally.
  • Error handling: Unknown errors are wrapped in n8n’s NodeOperationError so failures appear clearly readable in the UI.
  • Tracing in UI: The node uses addInputData/addOutputData so every call shows messages in and response/error out in the execution view.

STACKIT Embeddings

Use this node to generate vector embeddings from text.

Key options:

  • Model: dynamically loaded embeddings models (intfloat/* models)
  • Dimensions: choose target dimensionality if supported
  • Batch size and timeout controls
  • Strip new lines option for cleaner inputs

Typical usage:

  1. Add “STACKIT Embeddings” to the canvas (AI > Embeddings)
  2. Configure the node with your desired parameters (API-Key, Model and Options)
  3. Connect the output to vector stores, RAG flows, or custom logic

Behavior details:

  • Batch and timeouts: Inputs are batched (configurable) and timeouts accept seconds for small values (converted to ms). Newline stripping is enabled by default for cleaner inputs.
  • Error handling: Unknown errors are wrapped in NodeOperationError for readable failures in the UI.
  • Tracing in UI: The node logs inputs and outputs using addInputData/addOutputData so each request is traceable.

Compatibility

Compatible with n8n@1.60.0 or later.

Resources

Discussion