Search across languages without translating

Semantic and hybrid modes cross languages natively, so an English query retrieves relevant German passages; lang= narrows to one when you need it.

The problem

The usual approach is to translate the query into every target language and run N searches, which multiplies cost and quota by N and introduces translation error into the query — the one place you cannot afford it.

What to send

mode

semantic or hybrid

Only these cross languages. Keyword mode matches tokens and is monolingual by construction.

lang

Restricts results to one language, when you specifically want sources in it.

tld[]

de, fr

A different question — jurisdiction rather than language. Often what you actually meant.

The code

One query, any language backbash

# An English query retrieving German sources — no translation step.
curl -H "x-api-key: ulb_..." \
  "https://api.unlob.com/search?q=climate+policy+targets&mode=hybrid&lang=de"

The mistake to avoid

Using `mode=keyword` and wondering why nothing crosses languages. Keyword matching compares tokens, and the tokens are simply different — this is a property of the mode, not a coverage gap, and `why_not` will correctly tell you the document is present.

Frequently asked questions

Do I need to specify lang at all?

Only if you want to constrain the output. Leave it off and you get the most relevant passages in whatever language they were written; set it when your downstream consumer specifically needs one.

What about Chinese and Japanese?

Scripts without word delimiters are tokenised as character bigrams identically at index and query time, so they behave consistently in keyword mode as well as semantic.

Try it against your own queries

10,000 requests a month on the free tier, no card. Enough to run a real evaluation set rather than a demo.

Get an API key All recipes