EU-hosted inference for open-weight models
An OpenAI-compatible API for open-weight language models, served from dedicated GPU servers in the Germany or Finland and operated in Germany.
How we handle your data
- Prompts and outputs are processed in memory only. They are never written to disk or logs.
- Your data is never used to train or improve models.
- API traffic stays in the EU. GPU hosting: Hetzner Online GmbH (sub-processors).
- A GDPR data processing agreement applies to every customer: read the DPA.
- Request metadata (token counts, timings, status codes) is kept for 30 days for billing and abuse prevention.
Access
The service is available through inference marketplaces and directly to businesses. For direct API access or dedicated capacity, write to support@unwindly.app.