Sentiment Analyzer
100% private — runs on your device, never uploaded. Works offline once loaded.
Paste a review, comment or message and see whether its sentiment is positive or negative, with a confidence score. The AI model runs entirely on your device.
Private sentiment scoring in the browser
This tool runs a DistilBERT model fine-tuned on the SST-2 sentiment dataset, entirely on your device. It classifies text as positive or negative and returns a confidence score, so you can quickly gauge the tone of reviews, survey responses, support messages or social comments.
Nothing you paste is uploaded, which matters when the text is customer feedback, private messages or unpublished content.
Analyzing many items at once
Turn on “score each line separately” to classify a whole list of reviews or comments in one pass — each line gets its own label and confidence. This is handy for triaging feedback or spotting negative outliers.
Binary sentiment, and its limits
The model outputs a two-way positive/negative judgment; it does not detect a neutral class, sarcasm, or fine-grained emotion. Very short or ambiguous text can produce low-confidence or surprising results, so use the confidence score as a guide and read borderline cases yourself.
Frequently asked questions
Is my text uploaded?
No — analysis runs entirely in your browser and nothing is sent to a server.
Does it detect neutral sentiment?
No. This model is trained for a binary positive/negative decision, so neutral text is forced into whichever side scores higher — check the confidence for borderline cases.
Can it handle sarcasm?
Not reliably. Like most sentiment models, it can misread sarcasm or irony, which depend heavily on context.
What languages are supported?
The model is trained on English text and works best there.
Why is the first run slow?
A one-time model (~65 MB) downloads and caches on first use; later runs are fast.
Can I score a whole list at once?
Yes — enable “score each line separately” and put each item on its own line.
Advertisement