STACKIT AI Model Serving gives developers turnkey access to hosted open-source Large Language Models (e.g., Llama 3, Mistral, Qwen, Gemma) using OpenAI-standard REST APIs.
Service Overview
Section titled “Service Overview”- Sovereign GenAI: Fully European-hosted AI execution keeping sensitive prompt data private and GDPR-compliant.
- Cost Efficiency: Pay-per-token pricing models eliminate the high cost of maintaining idle GPU hardware.
- Drop-In Compatibility: Zero code refactoring for applications built on OpenAI API structures.
Technical Details
Section titled “Technical Details”- Deployment Options: Shared Model Serving (cost-optimized, multi-tenant) and Custom Model Serving (dedicated, predictable performance).
- Managed Scaling: Auto-scaling inference infrastructure managing underlying NVIDIA GPU resources natively.
Available Models
Section titled “Available Models”The list below comes from the STACKIT documentation and updates itself.
From the STACKIT docsAvailable Shared Models of STACKIT AI Model Serving
| Model | Capabilities | Context | Max. output | Status |
|---|---|---|---|---|
| Text models | ||||
| Qwen3-VL 235B | Chat, Vision | 200K | 16,384 | Supported |
| Qwen3.8 27B | Chat, Vision, Reasoning | 262K | 16,384 | Supported |
| Qwen3.6 27B | Chat, Vision, Reasoning | 262K | 16,384 | Deprecated |
| Llama 3.3 70B | Chat | 128K | 4,096 | Supported |
| GPT-OSS 120B | Chat, Reasoning | 131K | 8,192 | Supported |
| Gemma 3 27B | Chat, Vision | 131K | 4,096 | Deprecated |
| Gemma 4 31B | Chat, Vision, Reasoning | 256K | 4,096 | Supported |
| GPT-OSS 20B | Chat, Reasoning | 131K | 8,192 | Supported |
| Embedding models | ||||
| E5 Mistral 7B | Embedding | n/a | n/a | Supported |
| Qwen3 Vision-Language Embedding | Embedding | n/a | n/a | Supported |
What is this?
This section is copied from the STACKIT docs automatically, several times a day. It cannot be changed here. Changes belong in the STACKIT docs.
Limitations & Constraints
Section titled “Limitations & Constraints”- Coarse Token Permissions: API tokens grant access to all available shared models; fine-grained model-level access is unsupported.
- Stateless Endpoint: The API stores no session state or conversation context; context history must be sent with each request.
- Shared Cluster Variance: Shared tier latency fluctuates based on multi-tenant cluster demand and strict RPM/TPM rate limits.
Asset historyActive 2 of the last 12 weeksTMUpdatedNo updates · 1 bar = 1 week i
Maintainers
- ?Name not public?Name not publicThe Cloud Framework team knows who this is. The name is not shown on the site.
TM
Tobias M.Head of STACKIT Cloud Framework · STACKITOwnerActive 12 of the last 12 weeks · 168 updatesSTACKITwww.linkedin.com/in/tobias-müller-011304172??Name not publicThe Cloud Framework team knows who this is. The name is not shown on the site.Contributed in STACKIT
Tobias M.Head of STACKIT Cloud Framework · STACKITOwnerActive 12 of the last 12 weeks · 168 updatesSTACKITwww.linkedin.com/in/tobias-müller-011304172??Name not publicThe Cloud Framework team knows who this is. The name is not shown on the site.Contributed in STACKITTobias M.
Tobias M.Head of STACKIT Cloud Framework · STACKITOwnerActive 12 of the last 12 weeks · 168 updatesSTACKITwww.linkedin.com/in/tobias-müller-011304172
· Oct 2, 2026