Open-weight LLM inference on EU hardware. Zero data retention.

OpenAI-compatible API on our own GPU fleet in Romania. Qwen3.8 27B, 128k context, tool calls, streaming. Nothing you send is stored.

Request an API key Read the docs

EU only

Every request runs on cards we own, in three sites in Romania. No US provider in the path and no subprocessor for inference.

Zero data retention

Prompts and completions are never written to disk. We keep request metadata and token counts for billing, nothing else. Read the policy.

One price, no tiers

$0.30 per million input tokens, $1.80 per million output, cached input $0.08. Models and pricing.

Qwen3.8 27B

qwen/qwen3.8-27b on https://api.auxilio.ai/v1

128k context, 32k max output, FP8 weights on vLLM, own GPU hardware. Tool calls, JSON schema output, thinking on by default. Status: Serving.

Models and pricing

Drop-in for the OpenAI SDK

Point any OpenAI-compatible client at our base URL. Same request, same response shape, EU hardware.

from openai import OpenAI

client = OpenAI(base_url="https://api.auxilio.ai/v1", api_key="YOUR_KEY")
r = client.chat.completions.create(
    model="qwen/qwen3.8-27b",
    messages=[{"role": "user", "content": "Summarise GDPR article 28 in three lines."}],
)
print(r.choices[0].message.content)
Quickstart and docs

Need a key, or GPU servers by the hour?

API keys are issued within one business day. The same fleet is also available as virtual machines with one to eight GPUs.

Request an API key GPU Cloud
Offers valid for a limited time only. Other terms and conditions may apply. Click Here
Auxilio is not affiliated, associated, authorized, endorsed by, or in any way officially connected with NVidia Corporation.