Language Detection API

The Language Detection API covers 187 languages, identified by their standard ISO 639-3 codes. Every pricing tier includes all 187 languages — higher tiers only raise your monthly request quota and lower overage rates, they never restrict which languages you can detect.

When the input text is too short, too ambiguous, or mixes multiple languages, there isn't enough signal for a confident answer. Instead of guessing or throwing an error, the API returns the language code `und` (undetermined) and sets `reliable` to false. This is intentional so your pipeline always receives a well-formed JSON response it can branch on. Send longer or more representative text, lower the `min_length` threshold, or use the `only` constraint to improve results.

Use the GET endpoint for quick lookups and short strings that fit in a query string. Use the POST endpoint for long or multiline text, content with characters that are awkward to URL-encode, and any request where you want to constrain candidate languages with the `only` array. Both endpoints share identical detection logic and return the same response shape.

`min_length` maps to franc's minimum-length threshold. It defaults to 10 and is clamped between 1 and 1000. Text shorter than this threshold is reported as `und` with `reliable: false`. Lower it to attempt detection on very short snippets (accepting more undetermined results), or raise it if you only want confident answers from longer text.

The `only` parameter (available on the POST endpoint) accepts an array of ISO 639-3 codes that restricts detection to those candidate languages. If you already know your text belongs to a small known set — for example English, French, and German — constraining the candidates dramatically improves accuracy on short text because the detector no longer has to distinguish your languages from the other 184.

Detection runs entirely on Cloudflare's global edge network using lightweight trigram analysis with no model downloads and no cold starts. Detection itself completes in well under a millisecond, and because it executes close to your users at 300+ edge locations worldwide, end-to-end latency stays low everywhere. This makes the API suitable for real-time, high-volume pipelines.

Pass your API key as a query parameter: `?apikey=YOUR_KEY`. When you call the API through the RapidAPI marketplace, the proxy secret is injected automatically and you authenticate with your standard RapidAPI key headers instead. A missing or invalid key returns a 401 response.