European Inference
AI on our own hardware, in Europe.
Your prompts are never stored.
OpenAI-compatible · works with OpenCode, SDKs, curl
On-site. In Europe.
Every token is computed on machines we own and house in Europe. No hyperscaler detours, no transatlantic data flights — your requests enter and leave the continent exactly once.
EU jurisdiction
Hardware, operator and data paths under European law. GDPR is the default, not a clause.
Own hardware
Dedicated inference on our own on-site systems — not resold cloud slots, not brokered APIs.
Low latency for Europe
Served from within the continent: fast round-trips for European teams and agents.
Your prompts are never stored.
Content in, answer out, nothing written to disk in between. We meter tokens — nothing but tokens — to write your invoice.
Zero content retention
Prompts and completions are never logged, never stored, never used to train anything — by us or anyone else.
No analytics, no cookies
This very page ships no JavaScript and no trackers. Usage metering is token counts for billing — no profiling, no fingerprints.
Keys & TLS everywhere
Per-customer API keys, hashed at rest and revocable in one click. Let's Encrypt TLS on every connection.
Flexible by design.
Pick a plan that fits how you work — or tell us how you work and we'll build the plan around it.
Flat · Usage · Hybrid
Fixed monthly rate, pay-per-token, or a base fee with included quota. Switch whenever — invoices adapt.
Custom arrangements
Volume commitments, reserved capacity, specific models or isolation requirements: come work with us.
Start free
Free-tier models let you try the full pipeline — streaming, tools, agents — before spending a cent.
Drop-in for your stack.
One OpenAI-compatible endpoint. Streaming, tool calls and agent workloads included — point your client here and go.
{
"$schema": "https://opencode.ai/config.json",
"provider": {
"max": {
"npm": "@ai-sdk/openai-compatible",
"name": "European Inference",
"options": {
"baseURL": "https://slop.european-inference.de/v1",
"apiKey": "<your key — shown once>"
},
"models": { "fast": { "name": "Fast" } }
}
}
}Works with OpenAI SDKs, OpenCode, LiteLLM, anything that speaks /v1/chat/completions.