AI Models

One access, many language models.

A language model is the computing model behind every AI answer. No model is best at everything: some reply in seconds, others think long and carefully. At novendix.AI you therefore do not choose a provider once and for all, but the right model for each task – from 19 language models in three performance tiers, plus two automatic selectors that make the choice for you.

19 language modelsThree performance tiersSwitch models any timeNo training on your data

Four tiers instead of an endless list

The models are grouped by capability and cost. You do not need to know any model names to work well – but you can if you want to.

Smart Select

The automatic option. You simply start writing and the platform picks a suitable model based on your request. For everyone who would rather not deal with model names – and the sensible starting point.

Standard

Fast, economical models for everyday work: emails, summaries, translations, first drafts, simple analyses. Covers most of the daily workload and is included in every package.

Premium

More powerful models for more demanding tasks: longer documents, layered analyses, careful editorial work, more complex evaluations. Included from the Business package upwards.

Ultra

The strongest models available for the hardest cases: multi-step reasoning, extensive research, tasks where care matters more than speed. Also from the Business package upwards.

Which models are available

As of September 2026. Provider and context size are given, context size being how much text a model can take into account at once. The selection is extended continuously as new models meet our data protection requirements – the authoritative list is always the model list inside the platform. Changes to the line-up are tracked in AI News.

Automatik

Smart SelectPicks a suitable language model for your request
GPT Image Smart SelectPicks a suitable model for image generation

Standard – 10 models

Claude Haiku 4.5Anthropic · 200K context · Fast and inexpensive, for most chats
Gemini 3.5 Flash LiteGoogle · 1M context · Cheaper variant of Gemini 3.5 Flash
Gemini 2.5 FlashGoogle · 1M context · Inexpensive Gemini model
Gemini 2.5 Flash LiteGoogle · 1M context · Very inexpensive, for simple tasks
GPT 5.6 LunaOpenAI · 400K context · Standard GPT model
GPT-5 MiniOpenAI · 400K context · Compact GPT-5 for routine tasks
GPT-5 NanoOpenAI · 200K context · Fastest and cheapest GPT-5
o4-miniOpenAI · 200K context · Maths and reasoning
Mistral SmallMistral · 128K context · Fast Mistral model
DevstralMistral · 128K context · Code and development

Premium – 7 models

Claude Sonnet 5Anthropic · 1M context · Balanced everyday model
Gemini 3.8 FlashGoogle · 1M context · Fast Gemini model of the latest generation
Gemini 2.5 ProGoogle · 1M context · Strong reasoning model
GPT 5.6 TerraOpenAI · 400K context · For strategy and analysis
GPT-5OpenAI · 400K context · Versatile model for more complex tasks
Mistral LargeMistral · 128K context · Creative work and marketing
Mistral Medium 3.5Mistral · 256K context · The only Mistral with image recognition and reasoning

Ultra – 2 models

Claude Opus 5.5Anthropic · 1M context · Highest quality for complex tasks
GPT 5.6 SolOpenAI · 400K context · For structured data

Every model is marked with icons in the interface: a wrench stands for the use of tools such as web search or file access, an eye for image recognition.

Which model for which task?

A simple rule of thumb: the more independent thinking a task requires, the higher the tier. If in doubt, start with Smart Select.

Standard is enough

  • Drafting emails and form letters
  • Summarising or shortening texts
  • Translation and proofreading
  • Structuring minutes and notes
  • Generating simple tables and lists
  • Standard questions about stored documents

Premium or Ultra is worth it

  • Reviewing contracts and long documents thoroughly
  • Weighing several sources against each other
  • Preparing quotes and tender responses
  • Complex analyses with intermediate steps
  • Demanding code and architecture questions
  • Anything where a mistake gets expensive

You can switch models at any time using the model selector in the header – even in the middle of a conversation. The article Which language model for which task? covers the selection criteria in detail. The four model families are compared in Language models compared.

Why our model list is shorter

We only enable a model once it is contractually guaranteed that requests are neither used for training nor stored after the answer has been given.

  • Processing takes place in a German data centre.
  • Your inputs are not used to train AI models.
  • Requests are not stored at the model provider after they have been answered.
  • Processing on our behalf under Art. 28 GDPR is governed by contract.

What that means in detail is set out on the page Privacy & Security.

Frequently asked questions about AI models

What is a language model, exactly?

A language model is a computing model that has learned from very large amounts of text how language is put together. For every answer it predicts, word by word, what sensibly comes next. That creates the impression of a conversation. A language model knows nothing in the human sense and can be wrong – which is why every important answer needs to be checked by a person.

Which AI models are available at novendix.AI?

Twenty language models across the three tiers Standard (10), Premium (7) and Ultra (3), plus two automatic selectors for text and images. They include models from the Claude family by Anthropic, GPT by OpenAI, Gemini by Google and Mistral. The full list with provider and context size is further up this page.

Which model should I choose?

To begin with, Smart Select: the platform then chooses for you. If you want to steer deliberately, the rule of thumb is Standard for routine work, Premium for demanding texts and analyses, Ultra for tasks where care matters more than speed.

Can I switch models in the middle of a conversation?

Yes. The model selector sits in the chat header and can be changed at any time. The conversation so far is kept.

What does access to the models cost?

Access costs €25, €35 or €45 per user per month plus VAT. The Starter package contains the Standard models; Premium and Ultra are added from the Business package upwards. The details are on the page Pricing; which other cost blocks arise is worked through in What AI really costs in a company.

Are my inputs used for training?

No. Only models are enabled for which it is contractually guaranteed that requests are neither used for training nor stored after the answer has been given.

Try the models yourself for ten days

The best comparison is the one on your own task. In the trial access the models in your package are available immediately – no installation, straight in the browser.

Your secure AI platform for the Mittelstand. Secure. Intelligent. Integrated. Custom database integration, personally supported.

novendix GmbH · Industriestraße 6 · 91126 Schwabach
Locations: Schwabach · Weißenburg · Nuremberg
A company of the L&S Lange & Schermer Group

© 2026 novendix GmbH — All rights reserved.A New Era of Thinking · Built for the German Mittelstand 🇩🇪




novendix.AI is now available as an appDownload on the App StoreGet it on Google PlayDownload formacOSDownload forWindows