# HanseRouter

> HanseRouter routes requests through one OpenAI-compatible API to model providers hosted in the EU. One contract, one key, one invoice in euros. Closed beta in preparation.

Status: closed beta in preparation. The service is not publicly available yet. Request access at https://app.hanserouter.eu/registrieren or by email.

Website: https://hanserouter.eu/ (German), https://hanserouter.eu/en (English).

## What it is
- HanseRouter is an OpenAI-compatible API that routes your requests to model providers that process them in the EU.
- API base URL: https://api.hanserouter.eu/v1. Customer portal: https://app.hanserouter.eu.
- Existing OpenAI SDK code keeps working after swapping the base URL and key. Supported endpoints and models will be published at launch.

## Data protection
- Contracting party in Germany: Your contracting party is RN Edge Solutions GmbH, based in Schramberg, Germany.
- Operated on servers in Germany: The HanseRouter API and customer portal run on servers in Germany.
- Routing within the EU only: HanseRouter only routes requests to providers that process them in the EU. The provider processes the content to generate the response. Its terms apply in addition.
- No storage of content: HanseRouter does not store the content of your prompts or responses. Only usage counters for billing are stored, such as model and token count.
- Data processing agreement: We conclude a data processing agreement with you under Art. 28 GDPR.
- HanseRouter currently holds no certifications. Whether the service meets your legal requirements depends on your use case. We cannot make that assessment for you. We are happy to answer your questions about it.

## Providers (planned for launch, no contracts in place yet)
- Mistral (Based in France)
- IONOS (Based in Germany)
- STACKIT (Based in Germany)
- Scaleway (Based in France)
- OVHcloud (Based in France)
- Planned means: the integration is intended, contracts with the providers are not in place yet. The list may change before launch.
- Models from US providers: Not planned for launch. A later option may be to use them with your own key from the respective provider. Your contract with that provider then applies to those requests.

## Models in preparation (none enabled yet, no prices yet)
- Mistral AI: Mistral Large 4 (`mistralai/mistral-large-4`), up to 1M context, planned providers: Mistral. Large open model from Mistral for text and images. For demanding tasks with a very long context and tool calling.
- Qwen: Qwen3.8 27B (`qwen/qwen3.8-27b`), up to 262k context, planned providers: IONOS, Scaleway, STACKIT. Text and images in one model. For multi-step tasks, research and applications with tool calling.
- OpenAI: GPT-OSS 120B (`openai/gpt-oss-120b`), up to 131k context, planned providers: IONOS, Scaleway, STACKIT. Open model from OpenAI for demanding analysis, reasoning and workflows with AI agents.
- Mistral AI: Mistral Small 3.2 (`mistralai/mistral-small-3.2`), up to 128k context, planned providers: Scaleway. Compact multimodal model for multilingual chat, image understanding and structured tasks.
- Meta: Llama 3.3 70B (`meta/llama-3.3-70b`), up to 128k context, planned providers: IONOS, Scaleway, STACKIT. Versatile text model for assistants, content and multilingual dialogue. Supports the use of external tools.
- Qwen: Qwen3 Coder Next 80B (`qwen/qwen3-coder-next`), up to 256k context, planned providers: IONOS. For code generation, refactoring and debugging. Built for working with programming tools.
- Qwen: Qwen3.5 9B (`qwen/qwen3.5-9b`), up to 262k context, planned providers: IONOS. Small model for fast dialogue and image analysis. Combines a long context with tool calling.
- Qwen: Qwen3.6 35B A3B (`qwen/qwen3.6-35b-a3b`), up to 256k context, planned providers: Scaleway. For programming, reasoning tasks and agents. Processes text and images and supports tool calling.
- Google: Gemma 4 26B A4B (`google/gemma-4-26b-a4b`), up to 256k context, planned providers: Scaleway. Open multimodal model from Google for text and image applications with a long context.
- DeepSeek: DeepSeek V4 Flash (`deepseek/deepseek-v4-flash`), up to 256k context, planned providers: Scaleway. Open text and coding model with a long context.
- All models are in preparation and cannot be used through the API yet. Contracts with the providers are not in place yet and the selection may change. We will publish prices at launch.
- Catalogue: https://hanserouter.eu/en/models (English), https://hanserouter.eu/modelle (German).

## Pricing
- You pay the price of the respective provider plus a clearly stated markup for HanseRouter.
- Billing in euros: One invoice in euros from a German GmbH, no matter how many providers you use.
- Prepaid credit: You add credit in advance and usage is deducted from it.
- Markup shown openly: You see which part goes to the provider and which part to HanseRouter.
- The actual rates are not set yet. We will publish them at launch.

## Provider of this service
- RN Edge Solutions GmbH, Schramberg, Black Forest, Germany. VAT ID DE455108312.

## FAQ
- What is HanseRouter? HanseRouter is an OpenAI-compatible API. It routes your requests to model providers that process them in the EU. You have one contract, one key and one invoice in euros instead of billing with each provider separately.
- Can I use HanseRouter already? Not yet. The closed beta is in preparation. You can send us an email. We activate accounts one by one and will get back to you.
- Do I have to rewrite my code? No. Keep using the OpenAI SDK or a tool with an OpenAI integration and enter only the HanseRouter base URL and key. We will publish the endpoints and models supported at launch together with the provider list.
- Which providers and models are available? Planned for launch are Mistral, IONOS, STACKIT, Scaleway and OVHcloud. Contracts with these providers are not in place yet and the list may change. The model catalogue shows which models we are preparing. None is enabled yet.
- Does HanseRouter store my prompts? No. HanseRouter does not store the content of prompts or responses. Only usage counters for billing are stored. The provider of the selected model processes the content to generate the response. Its terms apply in addition.
- Does HanseRouter hold any certifications? No, not at this time. We tell you where the service is operated, where requests are routed and what we store. We conclude a data processing agreement with you.
- What does HanseRouter cost? You pay the provider price plus a clearly stated markup. Billing is in euros via credit that you add in advance. We will publish the actual rates at launch.
- Where does the name come from? The Hanseatic League was a European trading network led by German cities. HanseRouter routes requests between European providers. The service is built in Schramberg in the Black Forest.

## Contact
- noah.karrer@hanserouter.eu