SageMaker models
Text classification
Ready-to-deploy classifiers. Subscribe, create an endpoint or batch job, send text, get a label and confidence. Your data stays in your account.
What you get
A family of SageMaker model packages that all behave the same at the API. You are not signing up for a training platform — you pick the labels you need and deploy.
- Subscribe on AWS Marketplace, then deploy a real-time endpoint or a batch job in your account.
- Send JSON, CSV, or plain text; get a label and a confidence score back.
- Same invoke pattern on every model in the family — swap the package, keep your client code.
- Sample notebook on each product page so you can try a few examples after you subscribe.
The models
Each card is a separate Marketplace listing. Open it for samples, FAQs, and the notebook.
- SA
Sentiment Analysis Model
Classify sentiment of English text on SageMaker
Reviews, support tickets, and social listening.
SageMakerProduct page - CS
Content Safety Classifier
Moderate text for harm, NSFW, and bias on SageMaker
Moderation queues and user-generated content.
SageMakerProduct page - LI
Language Identifier
Detect which of 20 languages a text is written in
Language routing and i18n pre-processing.
SageMakerProduct page - ED
Emotion Detection Model
Detect the dominant emotion in English text on SageMaker
Surveys, chat transcripts, and product feedback.
SageMakerProduct page 
Financial Sentiment Analysis Model
Sentiment analysis for financial news and commentary on SageMaker
Earnings commentary, filings, and market news.
SageMakerProduct page
Formality Classifier
Classify English text as formal or informal on SageMaker
Tone checks for support replies and published copy.
SageMakerProduct page
Banking Intent Classifier
Classify banking queries into 77 intents on SageMaker
Banking chatbots and intent routing.
SageMakerProduct page
Spam Detector
Detect spam in messages and emails on SageMaker
Inbox, contact forms, and messaging pipelines.
SageMakerProduct page
News Topic Classifier
Classify news into World, Sports, Business, or Sci/Tech
Routing news and articles into a small topic set.
SageMakerProduct page
Sample result
A typical response is a label plus a score between 0 and 1. Exact labels depend on the model (positive/negative, spam/ham, a language code, and so on).
| Text | Label | Score |
|---|---|---|
| I love this product, it works great! | positive | 0.99 |
| This was a complete waste of money. | negative | 0.99 |
Looking for entities in text rather than a single label? See named entity recognition.