Aliru AI — LLM Inference

Open-weight language models, served on our own hardware in Germany.

Hosted in Falkenstein, Germanyno prompt retentionno training on customer data

Models

Model	Context	Input	Output
qwen3.5-122b-a10b	131,072	0.26 USD/M	2.08 USD/M
qwen3.5-27b	131,072	0.195 USD/M	1.56 USD/M
qwen3.5-4b	65,536	0.05 USD/M	0.20 USD/M

All models are open-weight (Apache-2.0) and are served unmodified.

API

The API is OpenAI-compatible. Changing the base URL is enough.

from openai import OpenAI

client = OpenAI(
    base_url="https://openrouter-api.aliru.ai/v1",
    api_key=API_KEY,
)
r = client.chat.completions.create(
    model="qwen3.5-122b-a10b",
    messages=[{"role": "user", "content": "Hello"}],
)

Streaming (server-sent events) and tool calling are supported. The model catalog is available at /openrouter/models.

Infrastructure

The models run on NVIDIA RTX PRO 6000 Blackwell GPUs in a data center in Falkenstein, Germany. No requests are forwarded to third parties.

Contact

Aliru GmbH, Julius-Hatry-Straße 1, 68163 Mannheim, Germany
kontakt@aliru.de · +49 621 49088670