Count tokens for the active note in the status bar, using the same tokenizers as popular LLM APIs (GPT, Claude, and Gemini). Desktop only.
- Open Settings → Community plugins.
- Turn on Community plugins and Browse.
- Search for Token Count and install.
- Enable the plugin.
- Download
main.jsandmanifest.jsonfrom the latest GitHub release. - Create a folder in your vault:
.obsidian/plugins/token-count/ - Copy both files into that folder.
- Enable Token Count under Settings → Community plugins.
Open any Markdown note. The status bar shows the token count for the full note, for example:
1,234 tokens · gpt-5
The count updates as you edit. It appears after Obsidian’s built-in word and character count when the core Word count plugin is enabled.
Settings → Token Count
| Option | Description |
|---|---|
| Model | gpt-5, gpt-4o, gpt-4, claude, or gemini |
| Show model in status bar | Show the model name after the count (e.g. · gpt-5) |
- Counts the entire editor content, including YAML frontmatter.
- Useful for estimating how much of a note fits in an LLM context window; system prompts or chat wrappers are not included.
- Gemini: the first count may download a vocabulary file once (requires network).
- Empty when no Markdown editor is active.
Requires Node.js 18+.
npm install
npm run buildSymlink or copy this repo into <Vault>/.obsidian/plugins/token-count/ for local testing. Use npm run dev to rebuild on save.
See RELEASING.md for publishing a new version.
MIT — see LICENSE.