How do you build Spanish AI apps with Plugsky?

Key facts

Spanish text	Latin script with ñ, accents and inverted punctuation (¿ ¡) handled as UTF-8
Address forms	tú, usted and vos vary by region and formality — pin one per surface
Regional variants	Spain and Latin America differ in vocabulary and verb usage; evaluate the variants you serve
Tokenisation	Long verb forms and accented text can split into subwords — measure with the token calculator
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

• Spanish runs on the standard endpoint with one base_url change.
• Pin tú, usted or vos per surface before writing prompts.
• Keep ñ, accents and inverted punctuation intact in display text.
• Separate Spain and Latin American evaluation slices.
• plugsky-embed-multilingual serves both variants 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 real Spanish prompts to /v1/chat/completions and compare two or three models.
3. Measure tokens with the token calculator, including accented and long verb forms.
4. Pick the address form and regional vocabulary per surface, and state them in the system prompt.
5. For RAG, embed with plugsky-embed-multilingual and test Spain and Latin American queries separately.
6. Score outputs on a Spanish gold set with native review before cutover.

Original data

Try it yourself

Open the token calculator →

How Plugsky handles Spanish text

Spanish needs no special endpoint: UTF-8 text goes in and text comes back, with ñ, accents and inverted punctuation preserved. The quality work is regional. Spain and Latin America differ in vocabulary — ordenador versus computadora, coche versus carro — and in how people are addressed.

Address forms are the most visible marker. tú, usted and, in parts of Central and South America, vos each come with distinct verb conjugations. If your product serves several markets, either localise per market or choose neutral phrasing and avoid forms that read as foreign anywhere.

Tokenisation and cost in Spanish

Spanish sits close to English in token cost. Accented vowels and ñ usually stay attached, but long verb forms and enclitic pronouns such as dámelo or decírselo can split into several subwords. Inverted punctuation marks are single characters, so they add little; the risk is stripping them during normalisation.

• Measure tokens on native Spanish, not translated English samples.
• Keep accents and ñ for display and embedding; fold for search keys only.
• Do not remove inverted punctuation before generation.
• Re-measure after model changes.

Spanish retrieval and RAG

One multilingual collection built with plugsky-embed-multilingual serves Spanish documents from either region with Spanish, English or mixed queries. Most retrieval errors come from regional vocabulary and accent folding rather than from the embedding model.

• Add accent-insensitive keys for keyword search alongside embeddings.
• Keep Spain and Latin American query sets separate during evaluation.
• Test English queries against Spanish documents explicitly.
• Keep one embedding model per collection.

Code example: a Spanish request

Point your OpenAI client at https://api.plugsky.com/v1; only the content and system prompt change.

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": "Responde en español latinoamericano con tratamiento de usted"}, {"role": "user", "content": "Resume este contrato en tres puntos"}])

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

Honest comparison

Capability	Plugsky	Spanish apps today	Building in-house
API compatibility	OpenAI-compatible — change base_url and model name	Varies by provider and SDK	Full rewrite
Regional coverage	One model with Spain and Latin American evaluation slices	Varies by provider and model	You collect and curate data
Token budget	Fixed tokeniser per model; measure native Spanish and chunk to fit	Varies by provider and model	You host and tune each tokeniser
Multilingual retrieval	plugsky-embed-multilingual for Spanish 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 Spanish text?

Yes. The API accepts UTF-8 Spanish input, including ñ, accents and inverted punctuation, on the OpenAI-compatible chat endpoint. Compare models on your own prompts.

Which Spanish variant does the model produce?

It follows your prompt and examples. State neutral Spanish, Spain Spanish or Latin American Spanish in the system prompt, and review outputs for regional vocabulary.

Should prompts use tú, usted or vos?

Match the market and formality: usted for formal service, tú for general consumer use, vos where it is locally standard. Keep one choice per surface.

How do I estimate Spanish token usage?

Use the token calculator on native Spanish text. Long verb forms and enclitic pronouns can split into subwords, so translated samples mislead.

Is there a multilingual embedding model?

Yes — plugsky-embed-multilingual is part of the 30+ model catalogue and serves Spain and Latin American content from one collection.

Can I keep data in my region?

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.