Two AI models.
German-language fine-tuning.
Operated in the EU.

For work where anonymization makes no sense – or simply isn't needed.

anymize Waterfall and anymize Fountain are our own AI models, built on Qwen 3.5, additionally fine-tuned on a curated German-language professional dataset and fully operated within the EU (at Hetzner in Germany). Available in chat, via API, with DPA and the consent statement under § 43e BRAO for German lawyers (equivalent professional-confidentiality frameworks apply across most European jurisdictions). For code, Excel, SQL, structured data – everywhere anonymization would destroy the data's value.

Explore the modelsStart for free

Why in-house models?

Not every task
tolerates anonymization.

For sensitive text the answer is clear: anonymize before sending to a frontier model. That's our second track – GDPR-compliant and powerful. But there are tasks where that route doesn't work:

Code

loses its structure when variable names, class identifiers or function signatures are swapped for placeholders.

Excel and CSV formulas

break when cell references or column names are anonymized.

JSON and API payloads

become inconsistent when keys disappear.

SQL queries

lose their semantics when table or column names turn into tokens.

For these cases there's only one clean solution: the model has to run where your data is allowed to sit anyway. In the EU, under a European company (GRVITY GmbH, Kiel, Germany), with the contracts you already signed with us.

That's exactly what anymize Waterfall and anymize Fountain are for.

The models

Waterfall for depth.
Fountain for speed.

Both models are Mixture-of-Experts architectures – they hold a large body of knowledge (many parameters), but activate only a fraction per request. That makes them markedly faster and more energy-efficient than comparable dense models, at equal or better quality.

Deep thinking

anymize Waterfall

Our strongest in-house model for deep analysis and multi-step reasoning.

PropertyValue
BaseQwen 3.5 122B-A10B (Mixture of Experts)
Parameters122 B total · 10 B active per token
Context window262,144 tokens native · up to 1 M via YaRN extension
Languagesover 200 supported, additionally fine-tuned on a German-language professional dataset
ModalityText · natively multimodal-capable
Usecomplex analyses, strategic planning, multi-step reasoning, legal and medical text work
Performance

anymize Fountain

Our performance model for high throughput and strong coding capabilities.

PropertyValue
BaseQwen 3.5 35B-A3B (Mixture of Experts)
Parameters35 B total · 3 B active per token
Context window262,144 tokens native · up to 1 M via YaRN extension
Languagesover 200 supported, additionally fine-tuned on a German-language professional dataset
ModalityText · natively multimodal-capable
Usewriting and refactoring code, Excel/CSV analyses, JSON and API processing, SQL work, structured data extraction

“One stack, two characters. Waterfall thinks deeply, Fountain delivers fast. Pick per question – or let anymize recommend.”

Both together

Note on currency

The base generation is Qwen 3.5. As soon as a new, stronger open-source generation appears (Qwen 3.6 Plus with 1 M context currently looks like the next step), we run another training pass – same name, higher quality.

German-language fine-tuning

What makes the models
stand out.

The Qwen 3.5 base is one of the strongest open-source models in the world. But: open-source models are optimized on English-language benchmarks. Continental European legal, medical and commercial vocabulary – and German contract, tax and labor law in particular – are underrepresented in the base models.

That's why we additionally train Waterfall and Fountain on a carefully curated German-language professional dataset:

Legal content

laws, judgments, pleadings, standard contract templates

Medical content

guidelines, specialist publications, treatment documentation

Commercial content

annual reports, economic reports, tax literature, business administration

All publicly accessible domain data, carefully curated – no synthetic data. The result: for German-language professional contexts (useful across Germany, Austria, Switzerland and any European team working with German-language material), our models are often more competitive than pure benchmark numbers suggest.

Performance positioning – honest

Open-source models are typically one version behind the proprietary frontier models from OpenAI, Anthropic or Google. Close – but not state of the art. Our edge: the German-language professional fine-tuning. For anyone analyzing German contracts, drafting German pleadings, structuring German findings, our answer is often the better one compared to an English-dominated frontier model.

And if you need the absolute top quality only frontier models deliver: that's what Track 1 is for – GPT, Claude, Gemini and friends, GDPR-compliant through automatic anonymization.

Where they fit

Where Waterfall and Fountain
shine.

Six scenarios from practice – with model recommendation:

Refactor code to internal standards

Fountain

Variable names, function signatures, comments stay intact – no anonymization

Excel analysis on company data

Fountain

Cell references, formulas, sheet names stay semantically untouched

Design SQL queries and optimize them

Fountain

Table/column names stay readable – test with real data

Structure a legal opinion

Waterfall

Strong German-language legal vocabulary + deep reasoning + no data outflow from the EU

Summarize a medical report

Waterfall

Strong German-language medical vocabulary + patient data without detour outside the EU

Strategic multi-step analysis on company data

Waterfall

Long documents, deep reasoning chains, fully internal

When Track 1 (international models) is the better fit: general-knowledge questions, creative text work, multimodal tasks, absolute frontier quality. That's where GPT, Claude, Gemini and friends play their strengths – and anymize automatically anonymizes for you before sending.

Access & legal framework

Direct access.
One contract.

Access

Waterfall and Fountain are available in two ways:

  • In chat — Selectable in the model dropdown next to the international models. One click per request.
  • Via anymize API — Callable like any other model. REST endpoints with API key authentication. Integrate them into your own automations, CRMs, agentic workflows (n8n, Make.com, Zapier, Flowise, MCP server).

The API lets you decide per request: with or without anonymization parameter. With Waterfall and Fountain, anonymization is not necessary – but of course possible if you want it as a second security layer.

Legal framework

Because inference, data storage and operation happen entirely with us, the contracts signed with you apply directly:

  • Data processing agreement (DPA) — under Art. 28 GDPR – automatically in force with terms acceptance, PDF available in settings at any time
  • BRAO consent statement — under § 43e BRAO (a German lawyer-confidentiality statute) – online text form for lawyers; written-form variant under § 26a BNotO for notaries. Equivalent professional-confidentiality frameworks apply across most European jurisdictions.
  • No third-country transfer — no data leaves the EU, no data reaches US providers, no US Cloud Act
  • No additional DPA — with a third party – you have the contract with us, not with OpenAI or Google
→ More on compliance controls
In preparation

anymize Spring

The little
sibling model.

anymize Spring is our third in-house model – currently in development. Significantly smaller than Waterfall and Fountain, but lightning-fast and resource-frugal. Two distinctives:

Ultra-fast answers

for real-time classification, simple text transformations, autocompletion, quick previews.

On-premise capable

Spring is offered as a model that can run in your own infrastructure. For organizations that have to host model inference themselves (banks, insurers, public sector with special requirements).

Waterfall and Fountain stay hosted with us in the EU. Spring extends our portfolio with the on-premise option for customers with maximum data-sovereignty requirements.

Available soon.

Who benefits

Who gets the most
out of it.

Engineering and dev teams

Code work without data outflow – embeddable into your own tools via API

Data teams

Excel/CSV/SQL analyses on company data, structured extraction, JSON transformations

IT decision makers

Clean contract chain, one vendor, no third-country transfer, predictable compliance

Compliance officers

No extra DPA, automatic DPA, consent statement under § 43e BRAO (for German lawyers; equivalent frameworks across EU), audit trail

Legal teams

Strong German-language legal vocabulary after fine-tuning (useful for teams working with DACH-region material), no anonymization needed for internal work

Healthcare / pharma

Strong German-language medical vocabulary, no data outflow from the EU, DPA + confidentiality included

And: every organization that wants to use AI productively, but wants to clear up the question "where does our data go?" once and for all.

What you should know about Waterfall and Fountain.

Frequently asked questions

Waterfall is our deep-thinking model: larger base (Qwen 3.5 122B-A10B), more active parameters per token, stronger at complex analyses and multi-step reasoning. Fountain is our performance model: smaller base (Qwen 3.5 35B-A3B), higher throughput, very strong coding capabilities. Both have a 262k native context window (extendable to 1 M), both are fine-tuned on a German-language professional dataset. Rule of thumb: Waterfall for deep analysis, Fountain for structured work at high pace.

Start now.
14 days free trial.

All models. All features. No credit card.

Start for freeHow it works

We stand behind anymize. And we know – when an AI tool touches client, patient or employee data, a demo video isn't enough. That's why we give you 14 days of full access – all models, all features, no credit card. Enough time to be certain, before you trust us.

Your AI workplace awaits.