← Retour au blog
tech 24 July 2026

Hetzner Delves into LLM Inference: A First Look

Hetzner is currently experimenting with large language model (LLM) inference using an OpenAI-compatible API model. Although still in the experimental stage, this initiative could change how businesses leverage LLMs on Hetzner's infrastructure.

Article inspired by the original source
Hetzner is working on LLM Inference ↗ sliplane.io

Introduction

Large language models (LLMs) are at the forefront of many recent AI innovations. Hetzner, known for its robust hosting solutions, is now exploring LLM inference. This article looks at Hetzner's ongoing experiment, its potential implications, and how it could redefine LLM usage.

What is Hetzner LLM Inference?

Hetzner Inference is an OpenAI-compatible API running on Hetzner's infrastructure. The concept is straightforward: allow users to test a language model without the usual complexity associated with managing computing resources. The current model, Qwen/Qwen3.6-35B-A3B-FP8, is a 35-billion-parameter Mixture-of-Experts model with 3 billion active parameters. It accepts text and images, has a 262K context window, and uses FP8-quantized weights.

Why Does It Matter?

The significance lies in the implications for tech businesses: accessibility and flexibility. Companies can test real workloads on an LLM model without immediately investing in expensive infrastructure. It's an opportunity to explore AI with minimal risk.

Hardware Challenges

Hardware is a crucial part of the equation. Currently, Hetzner does not offer a production-ready solution. There is no billing, SLA, or production guarantee. The goal is to learn how the system behaves, which features matter, and what kind of load it can handle.

Experiments and Initial Tests

The initial integrations with the API are straightforward thanks to OpenAI compatibility. For example, integrating this API requires just a few lines of Python code. A basic test might look like this:

```python pip install openai from openai import OpenAI

client = OpenAI( base_url="https://inference.hetzner.com/api/v1", api_key="YOUR_TOKEN", ) response = client.chat.completions.create( model="Qwen/Qwen3.6-35B-A3B-FP8", messages=[{"role": "user", "content": "Explain why the sky is blue in one sentence."}], ) print(response.choices[0].message.content) ```

Opportunities and Future

Should Hetzner decide to push this project beyond experimentation, it could offer companies a viable new option for LLM inference, particularly for those seeking to avoid the high costs associated with cloud giants.

Conclusion

Hetzner's initiative with LLM inference is a bold step that could redefine AI solutions for many companies. Although it remains an experiment, it deserves the attention of tech decision-makers. To discuss how this technology can integrate into your project, let's discuss your project in 15 minutes.

Hetzner LLM Inference OpenAI AI Infrastructure Mixture-of-Experts
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call