How do you build German AI apps with Plugsky?

Key facts

German script	Latin script with umlauts (ä, ö, ü) and ß; UTF-8 handled as-is
Compounds	Compound nouns split into many subwords — measure with the token calculator
Noun capitalisation	All nouns are capitalised, which increases variation in tokenisation
Register	Sie and du change pronouns and verb forms — keep one per product
Locale variants	de-DE, de-AT and de-CH differ in spelling, notably ß versus ss in Switzerland
API compatibility	OpenAI-compatible POST https://api.plugsky.com/v1/chat/completions — chat, streaming, JSON mode and function calling
Models	30+ models behind one endpoint, from free tiers to frontier reasoning
Free tier	Free plan with plugsky-micro and plugsky-lite, no card required

TL;DR

• German runs on the standard endpoint with one base_url change.
• Compounds and capitalised nouns push token counts above English.
• Pin Sie or du, and pick one national spelling variant.
• Test Swiss spelling if you serve de-CH (ß becomes ss).
• plugsky-embed-multilingual covers German and English retrieval.

How it works, step by step

1. Create a free Plugsky key and set base_url to https://api.plugsky.com/v1.
2. Send compound-heavy German text — legal, technical or HR — to /v1/chat/completions and compare models.
3. Measure tokens with the token calculator and add chunk headroom for compound-dense documents.
4. Decide Sie or du per surface and record the national spelling variant you support.
5. For RAG, embed with plugsky-embed-multilingual and test German and English queries against one collection.
6. Review outputs with native speakers on a fixed gold set, then move production traffic.

Original data

Try it yourself

Open the tokenizer →

How Plugsky handles German text

German needs no special endpoint: the API accepts UTF-8 text with umlauts and ß and returns the same. The language-specific work is mostly about form. German capitalises all nouns, joins compounds into single words and reflects formality through Sie or du across verbs and pronouns.

Spelling also varies by country. Germany and Austria use ß in words such as Straße, while Switzerland uses ss. Pick the variant your audience expects, state it in the system prompt and keep it consistent — mixed spelling reads as carelessness in formal content.

Tokenisation and cost in German

German usually costs more tokens than the English equivalent. Compounds such as Rechtsschutzversicherung or Donaudampfschifffahrt split into many subwords, capitalised nouns create extra case variants, and official written German favours long sentences with subordinate clauses.

• Measure on legal, insurance and technical text, where compounds cluster.
• Keep compounds intact for display; split for search keys only.
• Watch ß and umlaut forms if your pipeline folds characters.
• Re-measure whenever you switch models.

German retrieval and RAG

One multilingual collection built with plugsky-embed-multilingual serves German documents with German or English queries. Retrieval failures come mostly from compound variants, umlaut folding and the ß/ss difference between locales.

• Add compound-splitting or synonym keys for keyword search.
• Normalise umlauts and ß consistently before indexing, and keep originals for display.
• Test queries using the other locale's spelling.
• Keep one embedding model per collection.

Code example: a German request

Point your OpenAI client at https://api.plugsky.com/v1; the request shape matches any English call.

client = OpenAI(base_url="https://api.plugsky.com/v1", api_key=os.environ["PLUGSKY_API_KEY"])

client.chat.completions.create(model="plugsky-pro", messages=[{"role": "system", "content": "Antworte auf Deutsch, siezen Sie die Kundin und halten Sie die neue Rechtschreibung ein"}, {"role": "user", "content": "Fassen Sie diesen Vertrag in drei Punkten zusammen"}])

Start free with plugsky-micro and plugsky-lite, then compare paid models on a German gold set. See the docs for the API reference.

Honest comparison

Capability	Plugsky	German apps today	Building in-house
API compatibility	OpenAI-compatible — change base_url and model name	Varies by provider and SDK	Full rewrite
German text handling	UTF-8 with compound-aware chunking and Sie/du prompt control	Depends on provider tokeniser and prompt hygiene	You build normalisation and evals
Token budget	Fixed tokeniser per model; measure compounds and chunk to fit	Varies by provider and model	You host and tune each tokeniser
Multilingual retrieval	plugsky-embed-multilingual for German and English RAG	Often needs a separate embedding vendor	You serve and maintain embeddings
Sovereignty	Cloud, VPC, on-prem and air-gapped with residency options	Usually US/EU public endpoints	You own the full stack

Frequently asked questions

Can Plugsky handle German text?

Yes. The API accepts UTF-8 German input, including umlauts and ß, on the OpenAI-compatible chat endpoint. Compare two or three models on your own prompts before choosing.

How do German compounds affect token usage?

Joined compounds split into many subwords, so one word can cost several tokens. Measure on legal or technical text with the token calculator and leave context headroom.

Should prompts use Sie or du?

Sie for formal and B2B contexts, du for consumer and community products. State the choice in the system prompt and keep it consistent across templates.

Do I need to support Swiss spelling?

If you serve de-CH, expect ss instead of ß. Either generate per locale or choose wording that avoids the issue, and test both spellings in evaluation.

Is there a multilingual embedding model?

Yes — plugsky-embed-multilingual is part of the 30+ model catalogue and supports German and English retrieval from one collection.

Can I keep data in the EU?

Plugsky supports cloud, VPC, on-prem and air-gapped deployment with residency options; confirm specifics with the docs and the enterprise team.

Is there a free plan?

Yes — plugsky-micro and plugsky-lite are free with no card, and a 14-day full-access trial covers the paid catalogue.