A text shortener for word counts and character limits
Too many words for the form, the abstract, the post or the meta description? This text shortener cuts your text down to a target — a number of words, a number of characters, or a percentage of the original — and shows you whether the result fits.
You choose how. Keep my sentences uses no AI at all: it scores every sentence and keeps the best ones that fit, word for word, in their original order, so nothing can be misquoted. AI condense rewrites the text in fewer words with an AI model on your device (your browser’s built-in AI, or IBM’s open Granite 4.0, Apache-2.0), which reads more smoothly but has to be checked. For posts and search snippets, pick a platform such as X, Bluesky, LinkedIn or a meta description and the target is set for you.
How to shorten text
- Paste your text, or open a .txt, .md, .docx or .html file.
- Set a Target — words, characters or percent — or choose a platform under Fit to a platform limit.
- For Keep my sentences, choose where sentences are scored (on this device, or on our server with TypeSafe Jev), optionally tick “trim wordy phrases”, and press Shorten.
- For AI condense, press Condense with AI. If the result is still over the target, Trim to fit keeps only whole sentences that fit.
- Check the meter (words, characters and, where relevant, how X counts it), read the result, and copy or download it.
Text Shortener features
Three kinds of target
A word count, a character count, or a percentage of your text — the meter shows how much room is left, or how far over you are.
Platform limits
X 280 (links count as 23, emoji as 2), Bluesky 300, Threads and Mastodon 500, meta title 60 and description 160, LinkedIn headline 220 and post 3,000, Instagram 2,200, YouTube title 100, Google Ads 30/90.
No-AI extractive mode
Scores sentences by how central their words are to the whole text, with a bonus for opening sentences, and keeps the best ones that fit — never rewording them.
Optional server scoring
TypeSafe’s Jev model can score sentence importance instead; it is used only when you choose it and press Shorten, and the result is still your own sentences.
Wordy-phrase trimmer
“due to the fact that” → “because”, “are able to” → “can”, filler such as “basically” and “really” removed — 37 rules written for this page.
Keep my sentences or AI condense?
| Keep my sentences | AI condense | |
|---|---|---|
| How | Picks whole sentences that fit | Rewrites the text in fewer words |
| Wording | Exactly yours | New wording — read it |
| Can it invent facts? | No | It can drop or merge details |
| Hits the target exactly? | Never over; often a little under | Close, but may run over (then trim to fit) |
| Best for | Quotes, summaries of reports, legal or technical text | Posts, descriptions, abstracts you will edit |
| Where it runs | Your browser (or our server if you choose Jev scoring) | Your device |
How the sentence scores work
Each sentence is reduced to its main words (common words such as “the” and “with” are ignored, and simple endings like “-s” and “-ing” are trimmed). A sentence scores higher the more of its main words appear in other sentences — a sign it is about the text’s central topic — divided by the square root of its own word count, so long sentences are not favoured just for being long. The first sentence gets a bonus of 0.35, the first sentence of each paragraph 0.15, and sentences under five words lose 0.2.
The shortener then adds sentences from the highest score down, skipping any that would push the text over the target, and finally puts the chosen ones back in their original order. Open “Sentence scores” under the result to see every score and which sentences were kept. Because sentences are kept out of context, a pronoun such as “it” or a phrase like “the delay” can lose what it referred to — read the result once.
Character limits worth knowing
- X counts every link as 23 characters whatever its length, and most emoji and Chinese, Japanese or Korean characters as two. The meter shows the X count whenever it differs from the plain character count.
- Meta descriptions are cut by pixel width, roughly 150–160 characters on desktop and fewer on mobile; Google may also rewrite them. Keep the key point in the first 120 characters.
- LinkedIn and Instagram show only the first two or three lines before “see more”, so the opening sentence matters most. For SMS parts, bytes and more limits, use the Character Counter.
- Limits change. Those listed were checked in October 2026; leave a little margin.
Text Shortener FAQ
How do I shorten text to a word limit?
Paste it into this text shortener, choose Words, type the limit and press Shorten. The best whole sentences that fit are kept; for a smoother result, switch to AI condense and check the meter.
Does it send my text anywhere?
Not unless you choose “Server (TypeSafe Jev)” scoring and press Shorten — then the sentences are sent to this site’s server to be scored, and nothing is stored. Everything else, including AI condense, runs on your device.
Why is the result shorter than my target?
In “Keep my sentences” mode whole sentences are kept or dropped, so there is usually some room left. Try a slightly higher target, or AI condense.
Why does it say no sentence fits?
Every sentence on its own is longer than the target — common with a 30-character ad headline. Use AI condense, or write that one by hand.
Is this the same as a summarizer?
Close. The Text Summarizer gives you a summary in a format (bullets, a TL;DR); this page is about hitting a precise length and keeping as much of your own wording as possible.
Does the X count match X exactly?
It follows X’s published counting rules (links 23, emoji and CJK characters 2), which match X for ordinary posts. Paid accounts can post longer; the 280 limit applies to standard accounts.