Count tokens across text, JSON, Markdown, CSV, and code files without uploading their contents.
| File | Size | Characters | Tokens | Actions |
|---|---|---|---|---|
| Total | 0 B | 0 | 0 |
Your files stay in this browser. StackLens reads and tokenizes them locally.
Use the main tokenizer when you need to inspect and edit one piece of text directly.
Combine representative input and output totals with request volume and source-tracked model rates.
No. Supported text and code files are read and tokenized locally in your browser. File contents and file names are not sent to StackLens.
The first version supports plain text, Markdown, JSON, JSONL, CSV, HTML, XML, YAML, and common programming-language source files. Each file must be no larger than 2 MB.
Yes. Add up to 20 supported files and the table will show each file count plus combined characters, size, and tokens.
PDF and Word documents require text extraction before tokenization. Extraction order, headers, tables, and scanned pages can change the text, so StackLens does not treat those binary files as plain text.
No. The tool counts the decoded file text. Provider chat templates, system messages, tool schemas, media, and other request fields can add tokens.