Complete EU Hosting

All models in one place.

Use all models in Chat, Assistants, Agents, or via the API. GDPR-compliant hosted on European servers.

All listed models are immediately available and operated exclusively on EU servers. Your data never leaves the EU and is not used for AI model training.

Chat Models

Powerful language models for conversation and complex tasks

🇪🇺 🇩🇪

GLM 4.7

~300B

Powerful reasoning and coding model with long-context capability. Comparable to Claude Sonnet 4.5 and GPT-5.1.

Zhipu AI Open Source
reasoning long-context
🇪🇺 🇩🇪

Kimi K2 Thinking

MoE

Deep reasoning model with top performance on mathematical tasks. Achieves 100% on AIME 25.

Moonshot AI Open Source
deep-reasoning math
🇪🇺 🇩🇪

DeepSeek R1

671B MoE

Large reasoning model for deep analysis and complex problems. 671B MoE architecture.

DeepSeek Open Source
reasoning analysis
🇪🇺 🇩🇪

DeepSeek V3

671B MoE

High-performance 671B MoE model for complex tasks. Comparable to GPT-4o and Claude Sonnet 4.

DeepSeek Open Source
general complex
🇪🇺 🇩🇪

Qwen 3 235B

235B

Fast, powerful model with 262K context window. Ideal for long documents and complex analyses.

Alibaba Open Source
fast 262K context
🇪🇺 🇩🇪

GPT OSS 120B

120B

Open-source LLM with 120B parameters – high performance for versatile tasks.

OpenAI Open Source
general creative
🇪🇺 🇩🇪

Llama 3.3 70B

70B

Meta's latest Llama model, specifically optimized for instruction prompting.

Meta Open Source
general instruction
🇪🇺 🇩🇪

Llama 3.1 405B

405B

Largest Llama model (405B parameters) – maximum capacity for complex tasks.

Meta Open Source
complex analysis

Programming Models

Specialized in code generation, analysis, and explanation

🇪🇺 🇩🇪

Kimi K2.5 Coder

MoE

Natively multimodal coding model with SWE-Bench 76.8%. Understands code, images, and context.

Moonshot AI Open Source
code multimodal
🇪🇺 🇩🇪

Qwen 3 Coder 480B

480B MoE

Powerful open-source coder with SWE-Bench 67%. 480B MoE architecture for demanding code tasks.

Alibaba Open Source
code generation
🇪🇺 🇩🇪

GLM 4.7 Coder

~300B

SWE-Bench 73.8% – strong reasoning and coding model. Ideal for complex software projects.

Zhipu AI Open Source
code reasoning
🇪🇺 🇩🇪

GPT OSS 120B Coder

120B

Large 120B open-source model for code generation – powerful and precise.

OpenAI Open Source
code generation
🇪🇺 🇩🇪

Qwen 3 Coder 30B

30B

Qwen model trained for programming tasks – code generation and debugging.

Alibaba Open Source
code compact

Image Generation

Create images from text descriptions with state-of-the-art AI models

🇪🇺 🇩🇪

FLUX.1 [schnell]

12B

Lightning-fast image generation with impressive quality. Developed by Black Forest Labs.

Black Forest Labs Open Source
image fast
🇪🇺 🇩🇪

Stable Diffusion XL

6.6B

Stable and versatile image generation model. Ideal for creative and detailed images.

Stability AI Open Source
image creative

Audio Analysis

Transcribe and analyze audio content in over 99 languages

🇪🇺 🇩🇪

Whisper Large v3

1.5B

High-precision speech recognition in 99 languages. Perfect for transcriptions and subtitles.

OpenAI Open Source
transcription multilingual
🇪🇺 🇩🇪

Whisper Large v3 Turbo

809M

Fast speech recognition with optimized inference. Ideal for real-time applications.

OpenAI Open Source
transcription fast
Benchmark Comparison

How our models perform a direct comparison.

Our EU-hosted open-source models compared to leading proprietary models from OpenAI, Anthropic, and Google.

Our Models (Selection, Open Source, EU-hosted)

Model Type Intelligence Index SWE-Bench AIME 25 Tok/s Comparable to
GLM-4.7 🇪🇺
Reasoning 42 73.8% 95.7% 30 Claude Sonnet 4.5 / GPT-5.1
Kimi K2.5 🇪🇺
Multimodal + Agentic 47 76.8% 96.1% ~30 Claude Sonnet 4.5 / GPT-5.2
Kimi K2-Thinking 🇪🇺
Deep Reasoning 41 71.3% 100% ~46 o3 / Gemini 2.5 Pro
DeepSeek-R1-0528 🇪🇺
Deep Reasoning ~38 ~66% 87.5% 20 o3-mini / Gemini 2.5 Pro
DeepSeek-V3-0324 🇪🇺
General Purpose ~30 ~42% ~70% 25 GPT-4o / Claude Sonnet 4
Qwen3-Coder-480B 🇪🇺
Coding Specialist 25 67–70% 81.5% 55 Claude Sonnet 4 (Coding)

Showing a selection of our most powerful models. Over 15 models are available in total.

US Frontier (proprietary, for comparison)

Model Type Intelligence Index SWE-Bench AIME 25 Tok/s Comparable to
Gemini 3.1 Pro Reasoning 57 ~80% ~98% 110 Currently #1 overall
Claude Opus 4.6 Reasoning 53 80.8% ~99% ~50 #2–3 overall
Claude Sonnet 4.6 Reasoning 52 79.6% ~100% ~80 Best price/performance proprietary
GPT-5.2 (xhigh) Reasoning 51 ~78% 100% ~60 #3–4 overall
Claude Sonnet 4.5 Reasoning ~46 77.2% ~99% ~70 Predecessor, close to K2.5/GLM-4.7
GPT-5.3 Codex Coding 54 ~82% ~99% ~40 Best coder currently

Benchmarks based on publicly available results (SWE-Bench, AIME 2025, Intelligence Index). As of March 2026.

On Request

Premium models from leading providers

For special requirements, we also offer proprietary models. Contact us for a custom quote.

Anthropic
Anthropic
Claude Opus 4.6 200K
Claude Sonnet 4.6 200K
Claude Sonnet 4.5 200K
… and more
Request Quote
OpenAI
GPT-5.3 Codex Coding
GPT-5.2 128K
o3 Reasoning
… and more
Request Quote
Google
Gemini 3.1 Pro 2M
Gemini 2.5 Pro 1M
Gemini 2.0 Flash 1M
… and more
Request Quote
Connect your own models
Connect your own models
Platform without model costs
API connection to your infrastructure
Full control over your AI
Request Quote

These models require an individual agreement and separate billing.

Why Open Source?

Transparency meets performance

Open-source models offer decisive advantages over proprietary systems – especially for European businesses that value data sovereignty and flexibility.

Full transparency

The source code is openly accessible. You know exactly how the model works, what data it processes, and what results it produces. No black box, no hidden mechanisms – and the ability to independently verify how it works.

No vendor lock-in

With open-source models, you are not tied to a single provider. Switch flexibly between models depending on the task – without contractual restrictions, rising prices, or platform lock-in. Your workflows remain future-proof.

EU Hosting

Open-source models can be fully operated on European servers – completely independent of American companies and their legislation. This guarantees that your data stays in the EU and is subject exclusively to European data protection law – without risk from the US CLOUD Act.

Community-driven & Innovative

Open-source models benefit from a global developer community that continuously drives improvements and innovations. New capabilities and optimizations become available faster than with proprietary providers – and you benefit automatically.

Model guide

Which model fits your use case?

Lurus automatically selects the optimal model for your request. Here is an overview of which models are best suited for which tasks.

Use Case Recommended Model Why?
Marketing & copywriting GLM 4.7 Strong reasoning model with excellent text quality. Ideal for ad copy, social media posts, newsletters, and creative content.
Code & software development Kimi K2.5 Coder Our strongest coding model with SWE-Bench 76.8% – natively multimodal, understands code and context. For smaller tasks, Qwen 3 Coder 30B is used automatically.
Complex analyses & reasoning Kimi K2 Thinking Deep reasoning model with 100% on AIME 25. Ideal for mathematical problems, deep analyses, and logical conclusions.
Quick everyday questions Qwen 3 235B Fast, powerful model with 262K context window. Perfect for short questions, summaries, and daily tasks.
General complex tasks DeepSeek V3 High-performance 671B MoE model, comparable to GPT-4o. Ideal for demanding tasks, strategy development, and versatile use cases.
Multimodal analysis Kimi K2.5 Natively multimodal model – understands text, code, and images equally. Ideal for document analysis and visual tasks.
FAQ

Questions about our AI models

Data Sovereignty

Full control over your AI data.

European data sovereignty is not an option – it's the standard.

EU Data Centers

All AI models run exclusively in European data centers. Your data never leaves the EU – without exception. This supports your GDPR compliance and provides protection from the US CLOUD Act.

No Data Retention

After processing your request, all input data is immediately deleted. We store neither your prompts nor the generated responses on our processing servers. Only you decide where your chat histories are stored.

Contractually Assured

All data protection measures are contractually fixed. Our infrastructure meets high European data protection standards under GDPR. For enterprise customers, we create individual data protection agreements.

Certified Data Centers

Hosted exclusively in ISO 27001-certified data centers. Regular penetration tests and security audits ensure high standards. All connections are encrypted with TLS 1.3 and AES-256.

How It Works

How model switching works.

Switch seamlessly between AI models – with one click or automatically.

1

Enter your request

Ask your question as usual in the chat. Lurus automatically analyzes the complexity, context, and type of your request to select the optimal model.

2

Model is selected

Based on the analysis, Lurus selects the optimal model from your activated portfolio. For code tasks, a coding model is chosen; for analyses, a reasoning model; for images, a vision model.

3

Receive your answer

You receive the best possible answer – optimized for quality, speed, and cost. The model used is transparently displayed, and you can switch manually at any time.

Platform Advantages

Why businesses choose Lurus.

One platform, all models – these are the key advantages of the Lurus approach for your business.

Instead of committing to a single AI model, Lurus provides access to over 15 AI models through a single, GDPR-compliant platform. The key advantage: each model has different strengths.

🧠

Reasoning: GLM 4.7 and Kimi K2 Thinking excel at complex reasoning tasks.

💻

Coding: Kimi K2.5 Coder and Qwen 3 Coder 480B deliver top performance in software development.

📊

Analyse: DeepSeek V3 excels at complex analyses and data processing.

🎨

Image Generation: FLUX.1 creates high-quality images directly in chat.

Lurus automatically selects the optimal model for your request or you choose manually. All models run on European servers, completely independent of US corporations. Our models achieve benchmark scores on par with Claude Sonnet 4.5, GPT-5.2, and Gemini 2.5 Pro, at a fraction of the cost. Your data never leaves the EU, and all inputs are deleted immediately after processing. No hidden costs, no vendor lock-in, no compromises on quality or data privacy.

Industry Overview

AI models for every industry.

Regardless of your industry, our GDPR-compliant AI models offer tailored solutions – without compromising on data privacy.

Marketing & Communications

Create content, plan campaigns, and conduct market analyses – with models specially optimized for complex text processing. Confidential strategies are fully protected through EU hosting and immediate data deletion.

Sales & Customer Service

Create proposals, optimize sales materials, and efficiently handle customer inquiries. All models meet high data privacy requirements and are ideal for handling confidential business data under GDPR.

E-Commerce & Retail

Generate product descriptions, analyze market trends, and summarize internal reports. AI processes business data exclusively on EU servers – GDPR compliant.

Consulting & Advisory

Create market analyses, draft strategy papers, and optimize client presentations. Confidential client data is protected by EU hosting – no unauthorized third party has access to your consulting projects.

Ready for the future of AI?

Start now and use all models through a unified platform – GDPR-compliant.