How do you build Dutch AI apps with Plugsky?

Key facts

Dutch script	Latin script with compounds written as single words; no special encoding beyond UTF-8
Compounds	Long nouns such as arbeidsongeschiktheidsverzekering split into many subwords — measure with the token calculator
Register	Formal u and informal je change pronouns and verb forms — keep one per surface
Locale variants	Netherlands and Flemish vocabulary and phrasing differ; evaluate both if you serve both markets
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
Product status	Chat, streaming, JSON mode, function calling, embeddings, RAG and agents live; audio, images, moderation, files, batch, fine-tuning, assistants and responses coming soon

TL;DR

• Dutch runs on the standard endpoint with one base_url change.
• Compounds are the big token variable — measure legal, medical and HR text.
• Choose u or je per surface and enforce it in the system prompt.
• Split Netherlands and Flemish evaluation before launch.
• plugsky-embed-multilingual serves Dutch and English from one collection.

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 Dutch text to /v1/chat/completions and compare two or three models.
3. Measure tokens with the token calculator and add chunk headroom for compound-dense documents.
4. Write the u or je decision into your system prompts, per product surface.
5. For RAG, embed with plugsky-embed-multilingual and run Netherlands and Flemish query sets separately.
6. Review outputs with native speakers on a fixed gold set, then move production traffic.

Original data

Try it yourself

Open the token calculator →

How Plugsky handles Dutch text

Dutch is a Latin-script language and needs no special endpoint: send UTF-8 and the API returns text. The quality work is register and locale. Formal u and informal je change pronouns and verb forms throughout a sentence, so inconsistent prompts produce sentences that mix both.

Netherlands Dutch and Flemish Dutch differ in vocabulary, idiom and some phrasing. If you serve both markets, do not average them: pick vocabulary per locale, and where one product must serve both, choose wording that is neutral in both rather than correct in one.

Tokenisation and cost in Dutch

Word-for-word, Dutch costs about the same as English. Compounds break that symmetry: Dutch writes them joined, and a noun like arbeidsongeschiktheidsverzekering or ziekenhuisopname splits into several subwords. Legal, medical, HR and government text compounds heavily, so those documents deserve their own measurements.

• Count tokens on domain text, not on conversational samples.
• Keep compounds intact for display; add split keys for search.
• Watch hyphenated compounds and abbreviations in formal writing.
• Re-measure after any model change.

Dutch retrieval and RAG

One multilingual collection handles Dutch documents with Dutch or English queries when you embed with plugsky-embed-multilingual. Retrieval errors usually trace back to compound variants — the same concept written joined, hyphenated or split — and to locale vocabulary differences between Netherlands and Flemish users.

• Add compound-splitting or synonym keys for keyword search.
• Keep Netherlands and Flemish query sets separate during evaluation.
• Test English queries against Dutch documents explicitly.
• Store one embedding model per collection and re-embed on change.

Code example: a Dutch request

Point your OpenAI client at https://api.plugsky.com/v1; the request shape is identical to 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": "Antwoord in het Nederlands, gebruik u als aanspreekvorm"}, {"role": "user", "content": "Vat deze polis samen in drie punten"}])

Start free with plugsky-micro and plugsky-lite, then compare paid models on a Dutch gold set before cutover. See the docs for the full reference.

Honest comparison

Capability	Plugsky	Dutch apps today	Building in-house
API compatibility	OpenAI-compatible — change base_url and model name	Varies by provider and SDK	Full rewrite
Dutch text handling	UTF-8 with compound-aware chunking and u/je prompt control	Depends on provider tokeniser and prompt hygiene	You build normalisation and evals
Locale coverage	One model, with Netherlands and Flemish evaluation slices	Varies by provider and model	You collect and curate data
Multilingual retrieval	plugsky-embed-multilingual for Dutch 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 Dutch text?

Yes. The API accepts UTF-8 Dutch input on the OpenAI-compatible chat endpoint. Compare two or three models on your own prompts, since quality varies by task and domain.

How do compounds affect Dutch token counts?

Dutch joins compound nouns into one word, and tokenisers split them into several subwords. Use the token calculator on legal, medical or HR text, where compounds cluster.

Should prompts use u or je?

Match the surface: u for formal service, legal and enterprise flows; je for consumer and social products. State it in the system prompt and keep it consistent across templates.

Do I need separate handling for Flemish?

The API is the same, but vocabulary differs. Keep Netherlands and Flemish evaluation slices separate and avoid averaging the two in review.

Is there a multilingual embedding model?

Yes — plugsky-embed-multilingual is part of the 30+ model catalogue and supports Dutch 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 for evaluation.