---
title: "STACKIT AI Model Experiments"
description: "Managed MLflow hub for tracking machine learning experiments, logging parameters, and versioning model artifacts centrally."
scfAsset:
  category: "service"
  managed: true
  marketplaceUrl: "https://marketplace.stackit.cloud/en/products"
  tags: ["MLOps", "MLflow", "AI", "Data and AI", "Model Registry"]
  maintainers:
    - user: "alexander.gabert"
source_url: "https://framework.stackit.cloud/data-and-ai/assetcontainer/stackit/stackit-service-ai-model-experiments/"
source_file: "docs/data-and-ai/assetcontainer/stackit/stackit-service-ai-model-experiments.mdx"
---

STACKIT AI Model Experiments delivers a fully managed MLflow environment to track training runs, record metrics, and manage model registries without infrastructure overhead.

## Service Overview
- **Central MLOps Hub**: Eliminates fragmented local tracking setups by centralizing artifacts, prompts, and parameters.
- **Seamless Integrations**: Natively compatible with standard frameworks like PyTorch, TensorFlow, and LangChain.
- **Sovereign Execution**: Hosted in German data centers with full GDPR compliance and encrypted TLS connections.

## Technical Details
- **Architecture**: Managed MLflow service using STACKIT IAM/SSO for web UI access and token-based SDK/API authentication.
- **Storage Separation**: Metadata resides in a high-availability database; heavy model artifacts are stored in STACKIT Object Storage.
- **GenAI Tracing**: Full support for tracing LLM execution steps, prompts, and evaluation metrics.

## Storage and Access

The values below come from the STACKIT documentation and update themselves.

> From the STACKIT docs: [Architecture of AI Model Experiments › Artifact storage](https://docs.stackit.cloud/products/data-and-ai/ai-model-experiments/basics/architecture/#artifact-storage) (Source updated 07.04.2026, copied 05.10.2026)

Unlike the metadata, artifacts (models, images, large datasets) are stored within the user’s own STACKIT project space.

- Sovereignty: You retain ownership and control over your binary data.
- Access: The MLflow™ UI and SDK communicate directly with your storage bucket to upload and retrieve these files.

> From the STACKIT docs: [Architecture of AI Model Experiments › Authentication & security](https://docs.stackit.cloud/products/data-and-ai/ai-model-experiments/basics/architecture/#authentication--security) (Source updated 07.04.2026, copied 05.10.2026)

Security is managed through a combination of STACKIT permissions and application-level tokens:

- Management: Admins create and manage the instances and access tokens and set access roles via the STACKIT Portal or API.
- Logging data: Engineers use tokens to authenticate the Python SDK against the specific instance URI to log their training data.
- UI access: Each instance UI is accessible via a unique URI. Access is restricted via the STACKIT authentication and the roles of the user.

## Limitations & Constraints
- **Coarse Token Scoping**: Service tokens offer full workspace access; fine-grained read-only API roles are unavailable.
- **Project Scope**: Tracking instances and linked object storage buckets are strictly bound to a single STACKIT project.
- **No Native In-UI Backup**: Soft-deleted experiments are retained for 30 days before purge; point-in-time database restoration is not exposed in the UI.

<LinkCard title="STACKIT AI Model Experiments Documentation" href="https://docs.stackit.cloud" />