Languages

How do you build Swedish AI apps with Plugsky?

Swedish is Latin-script with å, ä and ö, and it builds compounds the way German does — with a twist: writers sometimes split compounds into separate words (särskrivning), which changes meaning. Plugsky serves Swedish on the standard OpenAI-compatible endpoint: set base_url to https://api.plugsky.com/v1 and post to /v1/chat/completions. Measure compound-dense text and use plugsky-embed-multilingual for Swedish and English retrieval.

Key facts

Swedish scriptLatin script with å, ä and ö; UTF-8 handled as-is
CompoundsCompound nouns split into several subwords; särskrivning changes meaning and evaluation
TokenisationCompound-heavy text inflates token counts — measure with the token calculator
Registerdu is standard in modern usage; ni as a polite form is unusual and easy to misread
API compatibilityOpenAI-compatible POST https://api.plugsky.com/v1/chat/completions — chat, streaming, JSON mode and function calling
Models30+ models behind one endpoint, from free tiers to frontier reasoning
Free tierFree plan with plugsky-micro and plugsky-lite, no card required
Product statusChat, streaming, JSON mode, function calling, embeddings, RAG and agents live; audio, images, moderation, files, batch, fine-tuning, assistants and responses coming soon

TL;DR

  • Swedish runs on the standard endpoint with one base_url change.
  • Watch compounds and särskrivning in both generation and search.
  • Use du; avoid ni as a polite form unless your brand insists.
  • Measure tokens on official and technical text.
  • plugsky-embed-multilingual covers Swedish 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 real Swedish text, including compound-heavy official prose, to /v1/chat/completions and compare models.
  3. Measure tokens with the token calculator and add headroom for compound-dense content.
  4. Set the du policy in the system prompt and keep templates consistent.
  5. For RAG, embed with plugsky-embed-multilingual and test Swedish and English queries against one collection.
  6. Score outputs on a Swedish gold set with native review before cutover.
1Create a freePlugsky key and setbase_url to2Send real Swedishtext, includingcompound-heavy3Measure tokens withthe tokencalculator and add4Set the du policyin the systemprompt and keep5For RAG, embed withplugsky-embed-multilingualand test Swedish6Score outputs on aSwedish gold setwith native review

Original data

Latin script wSwedish scriptOpenAI-compatiAPI compatibility30+ models behModelsSource: Plugsky facts table · updated 2026-09-26

Try it yourself

Open the prompt diff and evaluator →

How Plugsky handles Swedish text

Swedish uses the Latin alphabet with three additional letters, and the API handles it as plain UTF-8. The language-specific work concentrates in compounds and, to a lesser degree, address forms. Swedish joins words into compounds freely, and särskrivning — writing a compound as separate words, often under English influence — is a real quality issue that changes meaning rather than just style.

Modern Swedish uses du almost everywhere. The old polite ni can sound distant or even pointed in some contexts, so if your brand wants formality, get it from tone and wording rather than pronouns.

Tokenisation and cost in Swedish

Swedish writes compounds joined, and tokenisers split them into several subwords; official, legal and technical text compounds heavily. Number words and dates add further tokens in invoices and reports, and abbreviations are common in public-sector writing.

  • Measure on domain text, where compounds cluster.
  • Check that generation joins compounds correctly (no särskrivning).
  • Keep compounds intact for display; split for search keys only.
  • Re-count tokens whenever models change.

Swedish retrieval and RAG

A single multilingual collection built with plugsky-embed-multilingual serves Swedish documents with Swedish or English queries. Embeddings handle compound variants better than keyword search, but keyword indexes still need help: särskrivning and hyphenation produce different strings for the same concept.

  • Add compound-splitting or synonym keys for keyword search.
  • Test queries with both joined and split compounds.
  • Include English queries, common in technical teams.
  • Keep one embedding model per collection and re-embed on change.

Code example: a Swedish request

Point your OpenAI client at https://api.plugsky.com/v1; only the system prompt and content differ from an 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": "Svara på svenska, använd du-form och skriv ihop sammansatta ord"}, {"role": "user", "content": "Sammanfatta avtalet i tre punkter"}])

Start free with plugsky-micro and plugsky-lite, then test paid models on a Swedish gold set before moving production traffic. See the docs for the full API reference.

Honest comparison

CapabilityPlugskySwedish apps todayBuilding in-house
API compatibilityOpenAI-compatible — change base_url and model nameVaries by provider and SDKFull rewrite
Swedish text handlingUTF-8 with compound prompts and du-form controlDepends on provider tokeniser and prompt hygieneYou build normalisation and evals
Token budgetFixed tokeniser per model; measure compounds and chunk to fitVaries by provider and modelYou host and tune each tokeniser
Multilingual retrievalplugsky-embed-multilingual for Swedish and English RAGOften needs a separate embedding vendorYou serve and maintain embeddings
SovereigntyCloud, VPC, on-prem and air-gapped with residency optionsUsually US/EU public endpointsYou own the full stack

Frequently asked questions

Can Plugsky handle Swedish text?

Yes. The API accepts UTF-8 Swedish input, including å, ä and ö, on the OpenAI-compatible chat endpoint. Compare two or three models on your own prompts.

What is särskrivning and why does it matter?

Särskrivning is writing a compound as separate words, which can change meaning. Ask the model to join compounds in the system prompt, and test generation on known cases.

Should I use du or ni?

du is standard in modern Swedish. ni as a polite form is dated and can sound distant, so prefer du and express formality through tone and wording.

How do compounds affect token usage?

Joined compound nouns split into several subwords in the tokeniser, raising tokens per word. Measure on legal or technical text and leave context headroom.

Is there a multilingual embedding model?

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

Can I deploy in my own environment?

Yes. Plugsky supports cloud, VPC, on-prem and air-gapped deployment with residency options for regulated teams.

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.