Blink. Convert. Done.
Jupyter notebook to Markdown converter
Drop a Jupyter notebook below and get clean Markdown, with the token count and what it costs to send. It is converted in your browser — the file never leaves your device.
Rates checked 2026-08-31. See the cost on all 35 models → · Use this as context for a prompt →
What a IPYNB actually stores
A notebook is JSON with cells, which makes conversion exact rather than inferred. The judgement call is not how to read it but how much of the output to keep.
Asking a model to review, debug or explain an analysis works far better when it can see the code and the prose together in the order they were written. That ordering is the thing a notebook has and a bare script does not: the markdown cells carry the reasoning, and losing them turns a documented analysis back into undocumented code. Pasting the raw .ipynb instead is possible but wasteful - it is JSON, so every cell carries quoting, escaping and metadata you pay tokens for and the model gains nothing from. Converting first typically removes a large fraction of the file while leaving every line of code and every word of explanation intact.
What survives the conversion
Markdown cells as prose, code cells inside fenced blocks tagged with the kernel language so syntax highlighting and model comprehension both work, and cell order exactly as written.
What does not
Images and plots are dropped. Cell outputs are truncated past about 1,200 characters, because a notebook that printed a ten-thousand-row dataframe would otherwise bury the code you actually wanted to ask about.
Saying so plainly is deliberate. A converter that quietly drops half a document is worse than one that tells you which half, because you can work around a limit you know about.
Questions
Are cell outputs included?
Yes, but truncated. The code is usually the question and a huge printed dataframe is usually noise you would be paying tokens for.
Do plots come through?
No. They are images, and Markdown carries no image data a model can read here.
Which languages are supported?
Any kernel — the language is read from the notebook metadata and used to tag the code fences, so R, Julia and others work the same as Python.
Converting a different format
The same engine handles eighteen formats in total, all of them in your browser.
Token rates checked 2026-08-31. See the cost on all 35 models →
More free tools
All free, all instant, none of them need an account.