Your own labels, no training
Provide any set of labels and each one is scored against the text — change them whenever you like.
CatalogLatency Typically 1–3 seconds (scales with the number of labels) Async false Rate Limit Per API key Max Input Up to roughly 512 tokens of text
cn8/text-classify
Send a `text` and a list of candidate `labels`, and get back the best-matching label,
Learn more about Zero-Shot Classification Text NLPsyncInstant previewtextanalysisnlp
Try it
Text to classify
Candidate labelsType a label and press Enter.
sportspoliticstechnologyfinance
Result
Results show up here
The sample input is ready — just hit Run.
Overview
Send a text and a list of candidate labels, and get back the best-matching label, its confidence score, and an all_scores map of every label to its score. You define the labels at request time, so there's no training step and no model to manage. Responses are synchronous — one request, one result. Great for routing tickets, tagging content and detecting intent.
Key capabilities
Best label + full ranking
Get the single best label plus a complete label-to-score map for everything else.
One simple call
Send text and labels as JSON and get a structured result back.
Multilingual
Works across languages — keep your labels in the same language as the text for best results.
When to use it
Ticket routing
Sort support tickets into departments (billing / technical / sales) without a custom model.
Content tagging
Assign your own tags to articles, reviews or posts from any label set.
Intent detection
Detect user intent (greeting, question, complaint, request) from a label list.
Input & output
input
text: a string to classify, plus labels: an array of candidate labels.
JSON body
output
The top label, its score, and an all_scores map of every label to its score.
JSON
Guides & tips
How it works
- You provide a
textand a list of candidatelabels. - Each label is scored against the text, and the best one is returned at
data.labelwith itsscore. - The full ranking is available in
all_scores— sort it by value to see how every label scored.
Tips
- Send one text per request, along with the labels you want to score it against.
- Keep the label set focused (around 3–20 labels). More labels means a bit more processing time, but the price is the same.
- Use labels in the same language as the text.
- To rank labels yourself, sort
all_scoresby value (descending) — the top entry matchesdata.label. - If you have example texts for each category, use Few-Shot Classification instead.
Specs
Schema
Request body
textrequiredstringThe text to classify.
labelsrequiredarray<string>Candidate labels — keep them in the same language as the text for best results.
Response
labelstringBest-matching label.
scorenumberConfidence of the top label, from 0 to 1.
all_scoresobject<string,number>Map of every label to its score. Sort by value to rank the labels.
Pricing
Billed per request, regardless of the number of labels.
| Service | Unit | Price |
|---|---|---|
| Zero-Shot Classification | item | $0.006/request |