Powered by Hetzner Infrastructure
The Easiest Way to Run AI
on Hetzner GPUs
Don't manage bare metal drivers. SUPA runs on Hetzner GEX44 servers but gives you a serverless API. Deploy LLMs in minutes, not days.
The Problem
Hetzner GPUs are powerful, but setup is painful
- ✗ Hours spent configuring CUDA drivers and dependencies
- ✗ Ongoing server maintenance and security updates
- ✗ Managing model deployments, scaling, and failures
- ✗ Waiting weeks for GPU availability
The SUPA Solution
Serverless AI on Hetzner infrastructure
- No bare metal driver management
- No CUDA setup or GPU configuration
- No server maintenance overhead
- Instant API access to powerful models
- Pay-per-use pricing
- German data residency guaranteed
Infrastructure Specs
Enterprise-grade hardware, zero configuration
GPU
NVIDIA L4 (Hetzner GEX44)
Location
Falkenstein & Nuremberg, Germany
API
OpenAI-compatible
Latency
< 100ms TTFB
One API Call Away
Use the same OpenAI SDK you already know
curl https://api.supa.works/openai/v1/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer <YOUR_API_KEY>" -d '{
"model": "supa:instant",
"messages": [{"role": "user", "content": "Hello from Hetzner!"}]
}'Start running AI on Hetzner today
Free sandbox tier. No credit card required. German data residency guaranteed.
Get Your API KeyFrequently Asked Questions
Everything you need to know about running AI on Hetzner with SUPA
Running vLLM on Hetzner Cloud is challenging because it requires bare metal GPU servers with complex driver setup and ongoing maintenance. SUPA makes this serverless—we handle all the infrastructure, CUDA drivers, and vLLM deployment. You just make API calls and we run everything on Hetzner GEX44 servers for you.
Yes, Hetzner offers GPU instances through their auction-based system, but availability is limited and unpredictable. Managing these bare metal servers requires significant DevOps expertise. SUPA abstracts all of this—we secure GPU capacity on Hetzner infrastructure and provide you with a simple, serverless API that's always available.
OpenAI's GDPR compliance is complex—their servers are primarily in the US, which raises data residency concerns under GDPR. By using SUPA, your AI models run entirely on Hetzner servers in Germany (Falkenstein & Nuremberg), ensuring your data never leaves the EU and giving you a much safer alternative for GDPR-compliant AI hosting.