PDF.ai

PDF.ai now publishes four tiers, not two. Hobby is free forever and covers 1 PDF upload and 100 questions per month, OCR on 2 pages per file, and a 10MB file cap. Pro is $10/month billed yearly or $17/month billed monthly, raising that to 100 uploads, 1,000 questions, OCR on 10 pages per file, and 50MB files. Ultimate is $20/user/month billed yearly or $27/user/month billed monthly, and adds unlimited uploads and questions, the advanced AI model, OCR on 50 pages per file, and Capture & Ask. Enterprise is $30/user/month billed yearly or $37/user/month billed monthly, with 100MB files, OCR on 100 pages per file, team seats, and the white-labeled PDF embed.

Yes. PDF.ai includes built-in OCR so you can chat with scanned PDFs where the text exists only as images, old contracts, historical archives, photographed documents, and image-based exports all work. The OCR runs automatically on upload, so you don't need to pre-process files through a separate tool. Accuracy is solid on clean scans and weaker on handwritten or heavily distorted pages.

Capture & Ask lets you screenshot a specific area of a PDF (or any on-screen content) and ask PDF.ai a question about just that snippet. It's useful for asking about charts, tables, and diagrams that don't extract well as text, instead of pasting a mangled data dump, you give PDF.ai the visual directly. This feature depends on vision-capable models in the backend and is one of PDF.ai's most-used features for technical documents. Note the gating: PDF.ai lists Capture & Ask as available on the Ultimate plan or above, so it is not part of the free Hobby tier or the $10 Pro tier.

Yes, with one pricing caveat. PDF.ai lets you train a chatbot on one or more PDFs and embed it on your own site by adding a code snippet to your header. PDF.ai describes this Chatbot Widget as an additional paid plan available exclusively to Ultimate users and above, so it is an add-on rather than something bundled into the Ultimate subscription. Enterprise separately adds a white-labeled PDF embed. It's a common pattern for user manuals, FAQs, legal documents, and product knowledge bases where customers want to ask questions of a document instead of reading it. Combined with the API, this makes PDF.ai a realistic way to add document-grounded AI to a SaaS product without building your own RAG stack.

Yes, and it has grown into a proper API Hub with a Parse API for document parsing, an Extract API for data extraction, and a Chat API for conversational document queries. Commercial use runs through a paid API add-on billed in credits: $50/month for 1,000 credits, $112/month for 2,500, $200/month for 5,000, $350/month for 10,000, and a custom quote above that. Credits are consumed per action, at 2 for an upload, 1 for a GPT-3.5 message, and 4 for a GPT-4 message. Non-commercial use has its own separate limits. Responses return source citations so you can surface which page of a PDF an answer came from.

They're comparable products with slightly different positioning. ChatPDF is the older and more established player with strong chat quality and a larger free tier. PDF.ai is more developer-friendly with its API, embeddable chatbots, Capture & Ask, and browser extension. If you just want to chat with a few PDFs as a student, ChatPDF is a great default. If you want to build something on top of PDF chat, PDF.ai is the better choice.

Yes, and PDF.ai publishes the exact caps in its plan comparison. Hobby tops out at 10MB per file, Pro and Ultimate at 50MB, and Enterprise at 100MB. OCR coverage is tiered the same way, at 2 pages per file on Hobby, 10 on Pro, 50 on Ultimate, and 100 on Enterprise. For large files, PDF.ai chunks documents and uses RAG so the full content doesn't need to fit in a single context window, which makes it more robust than simply pasting a PDF into ChatGPT.

PDF.ai is built by an independent team and has grown through word-of-mouth in developer and research communities. It's not VC-backed at the scale of the biggest AI tools, which means pricing is relatively simple and the product prioritizes direct user requests over enterprise sales theater. The fact that a small team can ship a tool that competes with well-funded alternatives is a good sign of how commoditized PDF RAG has become in 2026.