DE
The right model per task
Chat, code, embedding and image: for each task you pick the model that fits – open or frontier, from twelve providers.
Models
Every model in one platform.
Open models such as Qwen, Mistral and gpt-oss run on your appliance, fully offline if you wish. Frontier models from OpenAI and Anthropic are yours EU hosted when your plan includes them.
- Local or fully offline
- Open and frontier models
- One interface for every task
Claude Sonnet 5Anthropic
Qwen3.6Available offline
GPT-5.6 SolOpenAI
MistralMistral AIModel catalogue
Mistral Large 3
- Context window
- 256,000 tokens
- Max output
- –
ChatEU hostedMistral Medium 3.5
- Context window
- 256,000 tokens
- Max output
- –
ChatEU hostedMistral Small 4
- Context window
- 256,000 tokens
- Max output
- –
ChatAvailable offlineMinistral 3 8B
- Context window
- 256,000 tokens
- Max output
- –
ChatAvailable offlineDevstral Small 2
- Context window
- 256,000 tokens
- Max output
- –
CodeAvailable offlineMagistral Small 1.2
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offlineCodestral
- Context window
- 256,000 tokens
- Max output
- –
CodeEU hostedMistral Embed
- Context window
- –
- Max output
- –
EmbeddingEU hosted
Qwen3.6-35B-A3B
- Context window
- 256,000 tokens
- Max output
- –
ChatAvailable offlineQwen3.6-27B
- Context window
- 256,000 tokens
- Max output
- –
ChatAvailable offlineQwen3-Coder-30B-A3B
- Context window
- 262,144 tokens
- Max output
- –
CodeAvailable offlineQwen3-30B-A3B-2507
- Context window
- 262,144 tokens
- Max output
- –
ChatAvailable offlineQwen3-14B
- Context window
- 32,768 tokens
- Max output
- –
ChatAvailable offlineQwen3-8B
- Context window
- 32,768 tokens
- Max output
- –
ChatAvailable offlineQwen3 Embedding 8B
- Context window
- –
- Max output
- –
EmbeddingAvailable offline
DeepSeek V4-Pro
- Context window
- 1,000,000 tokens
- Max output
- –
ChatGlobalDeepSeek V4-Flash
- Context window
- 1,000,000 tokens
- Max output
- –
ChatGlobal
Claude Fable 5.1
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedClaude Fable 5
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedClaude Opus 5.5
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedClaude Opus 5
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedClaude Opus 4.8
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedClaude Sonnet 5
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedClaude Haiku 4.5
- Context window
- 200,000 tokens
- Max output
- 64,000 tokens
ChatEU hosted
Gemini 3.1 Pro
- Context window
- 1,000,000 tokens
- Max output
- 64,000 tokens
ChatGlobalGemini 3.5 Flash
- Context window
- 1,000,000 tokens
- Max output
- 64,000 tokens
ChatGlobalGemini 3.1 Flash-Lite
- Context window
- 1,000,000 tokens
- Max output
- 64,000 tokens
ChatGlobalGemini Embedding 2
- Context window
- –
- Max output
- –
EmbeddingGlobalGemma 4
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offlineGemma 3 27B
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offline
GPT-5.6 Sol
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5.6 Terra
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5.6 Luna
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5.5
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5.4
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5.4 mini
- Context window
- 400,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5.1
- Context window
- 400,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT-5
- Context window
- 400,000 tokens
- Max output
- 128,000 tokens
ChatEU hostedGPT Image 2
- Context window
- –
- Max output
- –
ImageEU hostedgpt-oss-120b
- Context window
- 131,072 tokens
- Max output
- 131,072 tokens
ChatAvailable offlinegpt-oss-20b
- Context window
- 131,072 tokens
- Max output
- 131,072 tokens
ChatAvailable offline
Llama 4 Scout
- Context window
- 10,000,000 tokens
- Max output
- –
ChatAvailable offlineLlama 3.3 70B
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offlineLlama 3.1 8B
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offline
Phi-4
- Context window
- 16,384 tokens
- Max output
- –
ChatAvailable offlinePhi-4-reasoning-plus
- Context window
- 32,768 tokens
- Max output
- –
ChatAvailable offline
Llama 3.3 Nemotron Super 49B
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offlineNemotron Nano 9B v2
- Context window
- 128,000 tokens
- Max output
- –
ChatAvailable offline
GLM-5.2
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
ChatGlobal
Kimi K2.6
- Context window
- 256,000 tokens
- Max output
- –
ChatGlobal
MiniMax-M3
- Context window
- 1,000,000 tokens
- Max output
- –
ChatGlobal
51 of 51 models
Models marked available offline also run on your appliance, without internet. OpenAI, Anthropic and Mistral AI compute in EU tenants; global models only once approved per tenant. Which models you can use is set by your plan and your offer.
Missing a model?
We add further open models to your profile after a licence and quality review. Tell us which model you need.
Secure and sovereign
Run on your appliance, fully offline if you wish, or in European data centres.
Easy to integrate
Your applications talk to one stable interface. Which model answers is configured and logged on every request.
Transparent
Context window, maximum output and operating route are out in the open. An open model only changes its build once you approve an update.
Operating routes
First the data path, then the model.
A model name says nothing about where your request is processed. So you choose the operating route first; the models that fit follow from it.
Local on the appliance
Open models such as Qwen3.6, Mistral Small 4, Gemma 4, Llama and gpt-oss run on the DGX appliance in your building. No external model interface.
- Works without the internet
- No token costs
- A fixed monthly price
European On Demand
The same platform in European data centres, with no hardware of your own. Plus models from OpenAI, Anthropic and Mistral AI in EU tenants.
- GDPR data processing agreement
- A separate tenant
- Start for free
Global models
Models that run neither offline nor in the EU, such as Gemini, DeepSeek, GLM, Kimi and MiniMax. Once your administrator has enabled them, your users decide for themselves whether to use them.
- Enabled by your administrator
- Users decide for themselves
- Revocable at any time
Common questions
What people ask most before they choose a model.
Which model answers our questions?
You decide. On the appliance an open model such as Qwen3.6 or gpt-oss-120b does the work; which one depends on your tasks and how many people work at once, and is recorded in the offer. In European On Demand you can add frontier models.
Are our data used to train the models?
No. The models run as pinned versions and are not further trained on your content. What the platform learns from your documents sits as sourced knowledge in your knowledge layer, not in model weights.
Can we add a model that is not in the catalogue?
Yes, if it has open weights and a licence that allows commercial use. We review licence, quality and memory needs on the reference appliance and then add it to your profile.
Can we use GPT, Claude or Gemini models?
In European On Demand, yes, depending on the plan. OpenAI and Anthropic models run in the vendors' EU tenants; what goes there is the single request with selected excerpts, not your collection. Gemini computes globally; once your administrator has enabled it, users decide for themselves whether to use it. On the appliance, frontier models are an approved exception.
How do new model versions reach the appliance?
As reviewed packages. A new version goes through the same qualification as the first model and only replaces the old build once you approve it. Nothing is fetched automatically.
The right profile
Which models belong on your appliance?
Tell us your tasks and how many people work at once. We propose the model profile and record it in the offer.