AI Token Counter + Smart Text Chunker
Estimate tokens, check your context budget, and split long text into clean AI-ready chunks without uploading your content or using a paid API.
Add your text
Check your context budget
See how much room remains after reserving space for instructions and the AI response.
Your text fits comfortably within the selected planning budget.
Build clean AI-ready chunks
Choose a target size and let the tool preserve natural boundaries wherever possible.
Who this tool is for
This free tool is for content creators, writers, developers, researchers, students, and AI beginners who need to fit long text into ChatGPT, Gemini, Claude, or another AI assistant without guessing where to split it.
What you’ll have at the end
You’ll know the approximate size of your text, whether it fits your chosen context budget, and—when needed—you’ll have clean numbered chunks ready to paste into an AI conversation.
How to count tokens and split long text
- Paste your text or import a supported text file.
- Review the live token, word, character, sentence, and reading-time estimates.
- Check the context budget or open the Smart Chunker tab.
- Choose a target chunk size, overlap, and boundary method, then create and copy your chunks.
Quick win
Start with 2,000 estimated tokens per chunk and 150 tokens of overlap. That gives the AI enough context without creating giant, awkward blocks.
Level it up
Use Smart boundaries for articles and Markdown, Speaker turns for transcripts, and a larger overlap when names or instructions from one section must carry into the next.
Common mistakes and fixes
Mistake: treating an estimate as an exact model count
Fix: leave a safety margin. Tokenizers vary by provider, model, language, punctuation, and code.
Mistake: using zero overlap for connected material
Fix: add a modest overlap so the next chunk retains names, definitions, and the tail end of the previous idea.
Mistake: splitting only by character count
Fix: use Smart boundaries so headings, paragraphs, lists, speaker turns, and code blocks stay readable whenever possible.
Frequently asked questions
Does this show the exact token count?
No. It provides a planning estimate because each AI model may tokenize the same text differently. Exact counts require the matching provider tokenizer.
Is my text uploaded or saved?
No. Counting, file reading, and chunking happen locally in your browser. This module does not send the pasted text to the server or an AI API.
What files can I import?
The tool accepts common text-based files including TXT, Markdown, CSV, JSON, SRT, VTT, and HTML. It does not extract text from PDFs or Word documents.
What chunk size should I use?
Two thousand estimated tokens is a practical starting point. Increase or decrease it based on the AI platform, your response needs, and how much room you want to reserve.
Why add overlap between chunks?
Overlap repeats a small part of the previous chunk so names, definitions, and unfinished ideas are less likely to lose context.
