Why Uploading a .md File to AI Costs Less Than Uploading a .docx
You've done this before: drag a report into ChatGPT or Claude and ask it to summarize, rewrite, or answer questions about it. What you don't see is what happens in the half-second between "upload" and "reading."
That half-second is where the token cost gets decided — and it depends entirely on what kind of file you just handed over.
Uploading a .md file: nothing happens, because nothing needs to
A Markdown file is plain text. When you upload it, the AI reads it exactly as it is — the same characters you'd see in a text editor. There's no format to unpack first. What gets tokenized is your actual content, and nothing else.
Uploading a .docx file: it has to be figured out first
A .docx isn't text — it's a zipped bundle of XML files describing styles, formatting, and structure, wrapped around your content. Before an AI can read a single sentence, something has to open that archive, walk through the XML, and pull the actual words back out.
That extraction step is expensive in two different ways:
It can produce messier text than the original. Word splits sentences across multiple XML "runs" whenever formatting changes mid-sentence — a bolded word, a spell-check flag, an autocorrect — each carrying its own formatting metadata. Tables, headers, footers, and tracked changes add more of the same. Extraction tools do their best to reassemble clean text, but the result is often noisier than a file that was already just text.
Complex documents can get rendered as images instead of text. When a .docx (or PDF) has multi-column layouts, embedded tables, or heavy formatting, some AI platforms fall back to rendering pages as images and reading them visually rather than extracting text. Image tokens cost substantially more than text tokens for equivalent content — so a formatting-heavy Word doc can be one of the most expensive things you can upload, byte for byte.
A .md file never triggers either problem. There's no archive to unpack, no formatting to strip, and nothing to render as an image, because it was never anything other than text.
Why this matters more every month
AI file uploads aren't a novelty anymore — reviewing contracts, summarizing reports, chatting with documents are daily habits now, and every one has a token cost sitting underneath it. A file that reads cleanly on the first try is doing real, measurable work before the AI even responds.
This is exactly why keeping every document as a real .md file matters at DotMD. There's no format to unpack and nothing to extract, because the file the AI reads is the same file you were writing in — formatted visually with DotMD's toolbar the whole time, never converted, never rendered as an image just to be understood.
→ dotmd.co