ChatPDF, reviewed: the free-tier limit that's confirmed, the price that isn't, and which "ChatPDF" you're even using

What ChatPDF actually is

ChatPDF is a document-chat tool built by ChatPDF GmbH: upload a file, ask it questions, get answers with citations back to the page they came from. It launched years ago as one of the first "talk to your PDF" tools built on GPT, and its own site now describes its engine as dynamic routing between GPT-4o and GPT-4o-mini, picking whichever model fits the question. Beyond PDFs, it accepts Word documents (.doc/.docx), PowerPoint (.ppt/.pptx), Markdown, and plain text files, and it can hold several documents in one chat rather than just one at a time. It also markets itself as "best-in-class AI tools for students and researchers," with side tools for AI detection, an AI writer, flashcards, and slide generation bundled into the same interface.

Unlike a general chatbot, ChatPDF won't answer from thin air: it's built to ground its responses in whatever you uploaded, and it links its claims back to a source page. That's the same basic promise Gemini Notebook (formerly NotebookLM) makes, and it's the reason the two get compared constantly. The difference is scope: ChatPDF is built around one document, opened once, closed when you're done. Gemini Notebook is built around a growing collection you return to over weeks.

The free tier: one number is confirmed, the rest is folklore

ChatPDF's own FAQ states, in these words, that the free plan lets you "analyze 2 documents every day." That's the only free-tier limit we could find stated directly by ChatPDF itself. Nearly every third-party review repeats a fuller set of numbers: a 120-page cap, a 10MB file-size ceiling, and roughly 50 questions per day. These figures show up again and again across review-mill sites, consistently enough that they're probably close to right, but none of them trace back to anything on chatpdf.com. If you're planning around an exact page or file-size limit, test it with your actual document rather than trusting a number no primary source confirms.

What we could confirm directly from the site: it accepts documents in any language and lets you chat back in whichever language you prefer, it keeps a side-by-side view of your document and the chat, and it describes its citations as anchoring responses to PDF references, though nothing on the site specifies whether that's page-level or a more precise passage-level link.

The price nobody can agree on, including ChatPDF

This is the most useful thing we found for anyone about to hand over a card number. We went looking for ChatPDF's actual current price for its paid "Plus" plan and hit a wall three different ways. First, chatpdf.com/pricing returns a 404, as does chatpdf.com/privacy-policy: those exact URLs don't exist. Second, we pulled the site's raw HTML directly and searched it for dollar signs and price-adjacent text; the only "$" characters present are React server-component reference markers like $1 and $17, meaningless outside the page's internal rendering code, not prices. The actual Plus price is rendered client-side, likely inside a modal that opens only after you click "Plus upgrade" in the app, and we couldn't trigger that render from outside a full browser session.

Third, we checked what everyone else says, and they don't agree with each other. Across the review sites that rank for ChatPDF pricing queries, we found $4.99, $5, $6.99, $9.99, $11.99, and $19.99 a month, all presented as ChatPDF's current price. One Trustpilot reviewer, writing in January 2026, referred to "the €20+ monthly fee," a full currency and roughly double several of the USD figures, which may reflect real regional pricing or may just be one more inconsistent data point. The single number we could verify with a direct fetch, rather than a secondhand claim, comes from Apple's own App Store listing for the ChatPDF iOS app: an in-app purchase priced at $14.99 for the Monthly Plus Subscription and $89.99 for the Yearly Plus Plan (roughly $7.50/month if paid annually). That's Apple's own catalog metadata, not a blog's guess, which makes it the most trustworthy figure in this entire section. It's still not proof of the web price, since Apple's in-app purchase pricing routinely runs higher than a company's own website to absorb Apple's platform cut.

The honest takeaway: don't trust any single "ChatPDF costs $X/month" claim you read, including implicitly repeated ones on review sites that all cite each other. Open ChatPDF's own upgrade flow and read the number it shows you, in your currency, before you commit.

Which "ChatPDF" are you actually using?

This is the part almost nobody writing about ChatPDF mentions, and it matters because these are genuinely different products with different rules. The original is chatpdf.com, run by ChatPDF GmbH. But at least four other tools use the same or a near-identical name:

• [Monica AI's "ChatPDF"](https://monica.im/webapp/doc-chat), a completely different company. Its free tier gives registered users one query, caps files at 50MB and roughly 100,000 tokens (about 70,000 words), only accepts PDFs (no Word or PowerPoint), but explicitly states files will not be used to train the AI model, a clearer privacy statement than anything on the original ChatPDF's own site.
• QuillBot's "Chat PDF" tool, bundled inside the QuillBot suite alongside a separate, plain-text-only "Summarizer": three different tools with overlapping names under one brand, each with its own limits.
• NoteGPT's "ChatPDF" (notegpt.io/chat-pdf), a free browser-based version with its own separate quota structure.
• chatpdf.so, an unrelated site built on GPT-4 Turbo rather than GPT-4o.
• At least four to five distinct mobile apps in the Apple App Store and Google Play Store using "ChatPDF" or a near-identical name (one called "ChattyPDF AI / ChatPDF: AI Chat with any doc"), with no confirmed affiliation to ChatPDF GmbH.

If a friend recommends "ChatPDF" or you see it named in a listicle, don't assume it's chatpdf.com. Check the domain and the developer name before you upload anything, especially anything sensitive, since privacy policies differ across all of these and at least one (the original) doesn't clearly state its training-data policy at all.

What real users say went wrong

ChatPDF's Trustpilot page carries a 3.6-out-of-5 average, but on a sample of only nine reviews, small enough that a single unhappy customer swings the number meaningfully. The pattern in those reviews is bimodal rather than middling: roughly 44% are five stars, another 45% are one star, and almost nothing sits in between. Specific, named, dated complaints on the record there include:

• Nomvelo Dlamini (July 2025, 1 star) reported being charged after cancelling the subscription.
• A reviewer identified as CM (June 2024, 1 star) described the same pattern, calling it a scam via unauthorized post-cancellation charges.
• "Leapeppermint" (May 2024, 1 star) reported that failed upload attempts still counted against the daily free quota, burning through the day's allowance without ever getting a usable answer.
• "K K" (February 2025, 1 star) said a free upload simply didn't work and felt the marketing overstated what the free tier delivers.
• Heli Saari (January 2025, 2 stars) reported PDFs uploading out of order, unresponsive support, and called the price not worth it for what was delivered.

Against that, four five-star reviews clustered in early September 2025 ("Pablo," Vitta Fleysher, Kevin S, and Miro Liebschner) specifically praised the citation feature and multi-document support, and called it genuinely useful for study work, evidence the negative cluster isn't the whole picture. On the app stores, the official iOS app holds a stronger 4.6 out of 5 across 90 ratings, and the Android app reportedly holds a similarly strong rating across roughly 340,000 downloads (the precise ratings count varies by aggregator, so we won't repeat an exact figure we couldn't confirm directly on the Play Store listing itself), both meaningfully better than the small Trustpilot sample. Common Play Store complaint themes we found in aggregate (not individually re-quoted here, since we couldn't pull verbatim text from the store listing directly): the app crashing mid-session even for paying Plus users with no visible support contact, a paywall that appears before you can properly test whether the tool works for your document, and billing that continues after a user believes they've cancelled, echoing the Trustpilot pattern. One iPad-specific App Store review flagged that the tablet version is missing the text-selection and explain/summarize features that exist on desktop web, calling it, in their words, missing "the entire selling point" of the app on that platform.

The throughline across both the negative reviews and the aggregated complaint themes: if you subscribe, keep a record of your cancellation, and if a free upload fails, check your remaining quota before trying again.

What we couldn't verify, and won't pretend to know

In the interest of not padding this out with confident-sounding guesses: we found one secondary source claiming ChatPDF cannot process scanned or image-only PDFs (text-layer only, no OCR), but we didn't test this directly, and nothing on ChatPDF's own site addresses it either way. If your document is a photographed page or an old scan with no underlying text layer, assume it won't work until you've confirmed otherwise yourself. Separately, ChatPDF's site never states whether it uses your uploaded documents or chat content to train its models. It states documents are SSL-encrypted in transit and storage and that you can delete them anytime, but silence on training data is not the same as a policy, and we won't fill that gap with a guess. Compare that directly to Monica AI's competing "ChatPDF" tool, which states outright that it does not train on your files: on this one point, the copycat is more transparent than the original.

ChatPDF vs Gemini Notebook (NotebookLM): different jobs

These two get compared constantly because both promise grounded, cited answers instead of open-ended chat. But they're built for different-sized jobs, and the honest comparison is about scope, not which one is "better":

ChatPDF                              Gemini Notebook (NotebookLM)
-----------------------------------------------------------------------
Scope: one document at a time,        Scope: a growing source
  read once, close the tab              collection you return to
Free tier: 2 documents/day            Free tier: 50 sources per
  (confirmed); other limits              notebook, 500,000 words per
  commonly cited, not confirmed          source (both confirmed)
File types: PDF, Word, PPT,           Sources: PDFs, Docs, web links,
  Markdown, txt                          YouTube, audio, pasted text
Setup: no account required            Setup: Google account required
  to start a chat
Citations: page-level, claimed        Citations: page/passage-level,
  (granularity unconfirmed)              clickable, default-on
Audio/Video Overviews: no             Audio/Video Overviews: yes
Paid tier price: unverified as of     Paid tier: bundled into Google
  this writing (see above)               AI plans starting $4.99/mo

If you have one PDF you'll read once, ChatPDF (or a similar single-document tool) is genuinely lighter-weight: no notebook to set up, no other product to learn. If you're working across a growing pile of sources over weeks, want higher source caps, or want outputs like Audio Overviews and study guides, Gemini Notebook is built for that job and ChatPDF isn't trying to compete there. For a fuller lineup covering eleven document tools including both of these, our best AI PDF summarizer comparison breaks down OCR support, page limits, and citation quality side by side.

When ChatPDF is the wrong tool

• You need to work across many documents at once. ChatPDF is built around one document per conversation (multi-file chat exists, but it's not the tool's core design). A literature review or a research project spanning dozens of sources will outgrow it fast; Gemini Notebook's higher source caps are built for exactly that.
• Your document is a scan with no text layer. Unconfirmed but likely, per the one source we found: ChatPDF probably can't read it. Run OCR first, or use a tool that states its OCR handling explicitly, like Adobe Acrobat's AI Assistant.
• You need a firm answer on data training before you upload something sensitive. ChatPDF's own site doesn't say. If that matters for your document, either ask their support directly and get it in writing, or use a tool that states its policy outright.
• You're price-sensitive and need to know the exact cost before committing. Given how inconsistent third-party pricing claims are for this specific tool, don't budget around a number you read anywhere but ChatPDF's own upgrade screen, in your own currency, on the day you subscribe.

People also ask

Is ChatPDF free?

Yes, with limits. ChatPDF's own FAQ confirms the free plan lets you analyze 2 documents every day. Additional limits commonly cited elsewhere (120-page cap, 10MB file size, ~50 questions/day) appear consistently across review sites but aren't stated on ChatPDF's own site, so treat them as likely rather than confirmed.

How much does ChatPDF actually cost?

We couldn't verify the exact current web price: ChatPDF's pricing page 404s and the price isn't in the site's static HTML. Third-party sources disagree wildly ($4.99 to $19.99/month, one reviewer cited over €20/month). The one figure we could confirm directly is Apple's App Store listing: $14.99/month or $89.99/year for the iOS in-app purchase, which may not match the web price. Check ChatPDF's own upgrade screen before subscribing.

Is ChatPDF safe to use?

ChatPDF states documents are encrypted in transit and storage and that you can delete them anytime. It does not state anywhere on its site whether it trains its models on your uploaded documents or chat content, which is a real gap if you're uploading sensitive material. A competing tool that also uses the ChatPDF name, run by Monica AI, explicitly states it does not train on your files.

Does ChatPDF work on scanned PDFs?

Likely not, based on one secondary source stating it reads only the text layer of a document, not images. This wasn't independently tested and isn't addressed on ChatPDF's own site. If your PDF is a scan or photograph with no underlying text, run OCR on it first, or confirm with ChatPDF support before relying on it.

ChatPDF vs NotebookLM (Gemini Notebook): which should I use?

ChatPDF is lighter for one document you'll read once and close. Gemini Notebook is built for a growing source collection (50 to 600 sources depending on plan) you return to over time, and it adds outputs like Audio Overviews that ChatPDF doesn't have. They solve different-sized problems rather than competing head-to-head on the same job.

Are there multiple different ChatPDF apps?

Yes. chatpdf.com by ChatPDF GmbH is the original, but Monica AI, QuillBot, and NoteGPT each run their own separate tool using the same or a near-identical "ChatPDF" name, plus several unrelated mobile apps in the App Store and Play Store. Each has different limits and privacy policies, so check the domain or developer name before uploading anything sensitive.