Last month I cancelled my third paid PDF summarizer of the year. Every single one had the same problem: I was paying ten or fifteen dollars a month for something that free tools now do just as well, sometimes better.
Here is what I learned after testing eight free options for a 400-page research dump.
The honest free leaderboard
These are the tools I would actually recommend, ordered by how often I open them now.
1. NotebookLM (Google)
Free, no credit card, no daily limit on summaries. You drop in up to 50 sources per notebook (PDF, slides, websites) and ask questions across all of them at once. Citations come back as clickable footnotes. The German audio overview feature, which turns your sources into a two-voice podcast, finally shipped in German in May 2026 and works well for commuting through dense material.
Weakness: web only, no Outlook or Word integration.
2. ChatPDF
Free tier handles three PDFs per day up to 120 pages each. The fastest path if you just need to ask one question of one document. The chat box accepts German questions and answers in German if the source is German.
3. Claude Free / ChatGPT Free
Both handle PDF uploads in the free tier as of 2026. Claude is noticeably better at long-context summarization (200K tokens vs roughly 32K on ChatGPT Free). For anything over 100 pages, Claude is the right choice.
4. Local with PrivateGPT or Ollama
If you handle sensitive documents (contracts, medical, legal), nothing beats running a local model. PrivateGPT pipes your PDFs into Llama 3.1 8B locally. Slower, but the files never leave your machine, which makes GDPR-compliance trivial.
The workflow trap
The mistake I made for months was uploading one PDF, asking one question, then closing the tab. That treats AI like a search box.
What actually works: build a notebook around a topic, drop ten related PDFs in, then keep coming back to the same notebook over weeks. NotebookLM in particular gets more useful with more sources, because it can cross-reference them.
A note for German readers
I wrote a much longer version of this comparison in German, including a GDPR / Datenschutz section that English-speaking guides usually skip. If your documents are German contracts or invoices, the EU-hosting question matters: KI PDF zusammenfassen kostenlos: 8 Tools im Test 2026.
The DSGVO comparison there covers Auftragsverarbeitungsverträge for each tool, which one hosts in Frankfurt vs Virginia, and the three providers I would never feed a client document into.
What I actually use now
NotebookLM for research and study, ChatPDF for one-off questions, Claude for very long PDFs, local Ollama for anything I would not paste into a public chat.
Total monthly cost: zero euros. The first time I cancelled a 15-dollar subscription and replaced it with a free tool that worked better was genuinely irritating. Why did I pay for so long?
Probably because the paid tools market themselves and the free tools do not.
Top comments (0)