Drop-in OpenAI-compatible
Same request and response format as OpenAI. Point base_url at api.staik.se and your existing code, SDK, or agent works as-is — no rewrite.
Swedish GPU hardware
An LLM API that answers to Swedish law
OpenAI-compatible API on our own GPUs in Sweden. Swap your base_url, keep all your code. Real-time inference, no prompt content stored, no data leaves the country.
Swedish jurisdiction
All inference runs on our own hardware physically in Sweden. Beyond the reach of the US Cloud Act and free of Schrems II concerns. GDPR by design.
Nothing stored
Prompt content is processed in real time and deleted once the response is delivered. Your data never trains our models and is never shared with third parties.
Modern open models
The Qwen3 family and more open-weight models on 2× RTX 3090 and up. Long context, function calling, and streaming — ready for agentic coding and RAG.
FAQ
How do I switch from OpenAI?
Set base_url to https://api.staik.se/v1 and use your staik key. The rest of your code is unchanged because the API is OpenAI-compatible.
Where do the models run?
On our own GPU hardware in Sweden. No calls go to US clouds and no data leaves the country.
Are my prompts stored?
No. Prompt content is not stored after the response is delivered and is never used for training.
Which models are available?
Open models such as the Qwen3 family, with long context and function calling. The current list is in the docs.
Build on an API that stays in Sweden
Create a key and make your first call in minutes.
Open api.staik.se