An OpenAI-compatible API to our self-hosted models: Qwen, MiniMax-M2, German embeddings, Whisper and Higgs TTS. A fixed base fee with included volume, guaranteed capacity on our own GPUs in Germany, DPA and German support included.
Why with us
Sovereign, not interchangeable
No US subprocessorInference on our own GPU servers in Germany, DPA included.
OpenAI-compatibleSwap the base URL and you are done. SDKs and tools keep working.
No anonymous token discountNamed contacts, onboarding and prompt help are part of the plan.
Frequently asked
Questions about this offer
Why a base fee instead of raw token prices?
Because you get more than anonymous tokens: guaranteed capacity on our own GPUs in Germany, a DPA, German support and named contacts. Raw token discounters do not fund that.
Do you train on our data?
No. Prompts and completions are not used for training and are not retained after processing unless you want logging.
Is the API really OpenAI-compatible?
Yes, /v1/chat/completions and /v1/embeddings. Usually swapping base URL and API key is all it takes.
We use strictly necessary cookies and, only with your consent, Google Analytics for anonymous visit statistics. No advertising trackers.
More in the cookie policy
