IONOS CLOUD AI Model Hub | LLM Hosting
AI inference on European infrastructure
- Operated and processed exclusively in European data centres – fully GDPR-compliant
- No extraterritorial data-access obligations – your operations are governed solely by European law
- OpenAI-compatible REST endpoint – integrate existing codebases and frameworks directly
Sovereign AI infrastructure built for teams that can't compromise on compliance
Data sovereignty built in
All inference requests are processed in European data centres and discarded immediately afterwards – sovereignty is structural, not a contractual add-on.
- No US CLOUD Act exposure, no extraterritorial disclosure risk
- No third-party access – strict data sovereignty
- Your prompts and outputs are never used to train models
Fully managed inference layer
Your teams write against a stable REST API while IONOS CLOUD handles the entire serving stack, model versioning, and availability.
- No GPU infrastructure to procure, configure, or maintain
- Model lifecycle managed by IONOS CLOUD, not your engineering team
- Scales from first test to production on the same API
Seamless, OpenAI-compatible integration
Native OpenAI-compatible REST endpoints mean existing tools, SDKs, and frameworks integrate without modification.
- Works with the official OpenAI SDKs and any HTTP client
- Chat completions, image generation, and embeddings on one API
- Playground in the DCD for model testing
Retrieval augmented generation
Combine large language models with your own document collections to produce responses that are accurate, contextual, and grounded in your data.
- Pair embedding inference with the IONOS CLOUD Managed Database (PostgreSQL + pgvector) – or a vector store of your choice
- Embedding and LLM inference on the same sovereign platform
- Multilingual embedding models for broader, cross-lingual coverage
Function calling and structured outputs
Models return schema-valid JSON and structured tool-call requests your application executes. This means clean integration into downstream systems, with no fragile text parsing.
- Trigger workflows and retrieve real-time data via natural language
- Integrate directly with business applications, no custom middleware required
You pay only for the tokens you consume – no standing charges, no minimum spend, and no cost when workloads are idle.
- No upfront investment — no hardware to buy or capacity to reserve
- Input and output tokens billed separately per model
- Transparent usage data supports internal budget governance
Pay only for what you actually use
Model | Price per 1 million input tokens | Price per 1 million output tokens |
|---|---|---|
Llama 3.1 8B Instruct
| £0.14 | £0.14 |
Mistral Nemo Instruct | £0.14 | £0.14 |
Mistral Small 24B Instruct | £0.085 | £0.26 |
gpt-oss-120b | £0.14 | £0.55 |
Llama 3.3 70B Instruct
| £0.55 | £0.55 |
Llama 3.1 405B Instruct | £1.58 | £1.58 |
Llama 3.1 8B Instruct
| |
|---|---|
Model | |
Price per 1 million input tokens | £0.14 |
Price per 1 million output tokens | £0.14 |
Mistral Nemo Instruct | |
Model | |
Price per 1 million input tokens | £0.14 |
Price per 1 million output tokens | £0.14 |
Mistral Small 24B Instruct | |
Model | |
Price per 1 million input tokens | £0.085 |
Price per 1 million output tokens | £0.26 |
gpt-oss-120b | |
Model | |
Price per 1 million input tokens | £0.14 |
Price per 1 million output tokens | £0.55 |
Llama 3.3 70B Instruct
| |
Model | |
Price per 1 million input tokens | £0.55 |
Price per 1 million output tokens | £0.55 |
Llama 3.1 405B Instruct | |
Model | |
Price per 1 million input tokens | £1.58 |
Price per 1 million output tokens | £1.58 |
Model | Price per 1 million input tokens | Price per 1 million output tokens |
|---|---|---|
LightOnOCR 2 New | £0.13 | £0.26 |
LightOnOCR 2 New | |
|---|---|
Model | |
Price per 1 million input tokens | £0.13 |
Price per 1 million output tokens | £0.26 |
Price | |
|---|---|
FLUX.1 [schnell] | £0.0256 per image |
FLUX.2 [klein] New | £0.011 per megapixel (first) |
FLUX.2 [klein] New | £0.001 per megapixel (additional) |
FLUX.2 [klein] New | £0.001 per megapixel (input) |
FLUX.1 [schnell] | |
|---|---|
Price | £0.0256 per image |
FLUX.2 [klein] New | |
Price | £0.011 per megapixel (first) |
FLUX.2 [klein] New | |
Price | £0.001 per megapixel (additional) |
FLUX.2 [klein] New | |
Price | £0.001 per megapixel (input) |
Price per 1 million tokens | |
|---|---|
paraphrase-multilingual-mpnet-base-v2 | £0.01 |
bge-large-en-v1.5 | £0.015 |
bge-m3 | £0.02 |
Qwen3-VL Embedding-8B New | £0.09 |
Qwen3-VL Reranker-8B New | £0.034 |
paraphrase-multilingual-mpnet-base-v2 | |
|---|---|
Price per 1 million tokens | £0.01 |
bge-large-en-v1.5 | |
Price per 1 million tokens | £0.015 |
bge-m3 | |
Price per 1 million tokens | £0.02 |
Qwen3-VL Embedding-8B New | |
Price per 1 million tokens | £0.09 |
Qwen3-VL Reranker-8B New | |
Price per 1 million tokens | £0.034 |
Models | Price per 1 million input tokens | Price per 1 million output tokens |
|---|---|---|
Qwen3-Coder-Next (80B) New | £0.13 | £0.68 |
Qwen3-Coder-Next (80B) New | |
|---|---|
Models | |
Price per 1 million input tokens | £0.13 |
Price per 1 million output tokens | £0.68 |
What your teams can build with IONOS CLOUD
Use case
Use case
Sovereign AI assistant for internal services
Develop internal or citizen-facing assistants on an inference layer that structurally meets data-protection requirements – no exceptions, no trade-offs against usability.
- Processing sensitive requests without data leaving the EU
- Chat Completions API as the foundation for customised conversational logic
- Your prompts and outputs are never used to train models
- Deployable where other providers are ruled out on data protection grounds
Use case
Use case
Search and retrieval across internal data
Create RAG pipelines that unlock your internal document inventory. IONOS CLOUD provides the embedding and LLM inference, your documents never leave the platform.
- Pair embedding inference with the IONOS CLOUD Managed Database (PostgreSQL + pgvector), or a vector store of your choice
- Semantic similarity search across confidential internal holdings
- Context-rich responses through the combination of retrieval and inference
Use case
Use case
Automated contract and document processing
Automate the handling of incoming contracts, reports, and correspondence. Structured extraction of relevant fields and direct handover to ERP and DMS systems, without manual intermediate steps.
- Models deliver structured outputs – results flow directly into ERP, DMS, or specialist applications
- Open-weights basis: predictable model behaviour, auditable and reproducible
- Open, standardised API — no vendor lock-in, no proprietary format
Documentation and guides
Getting started
API interface
Enrich your apps
- Register for the IONOS CLOUD Public Cloud
- Get full access to the IONOS CLOUD AI Model Hub API
- Connect powerful AI models to your application
The IONOS CLOUD in practice
A catalogue of open-weight AI models
Generate high-quality content in seconds with an open-source language model from our AI Model Hub.
- Texts written like a native
- Creative writing and storytelling
- Code generation
Reduce lengthy articles, paragraphs, or documents to their essential points with LLMs.
- Summaries
- Abstracts
- Briefs
Use LLMs to select similar texts based on content or context and categorise them effectively.
- Sentiment analysis
- Topic labelling
- Tagging content
With the AI Model Hub, you can choose relevant texts based on predefined criteria and fine-tune the accuracy of search results.
- Implement a relevance ranking
- Enhanced query understanding with LLMs
Store and process text elements more efficiently than with SQL or NoSQL databases using a vector database.
- Identify relevant content for queries and searches
- Provide content to an LLM for processing
Benefit from precise and customised image generation with AI, creating unique visuals in no time.
- Convert existing texts into visual representations
- Create images based on text inputs
Our product experts are here from 9:00–17:00, Monday to Friday.
