100 free credits · No card required

PDFs for developers. Not headaches.

Generate, merge, split, watermark, extract. Every PDF operation is one API call. Replace five libraries with one endpoint.

Get 100 free credits Read the docs

TypeScript + Hono Open source SDK Sub-200ms response

$ npm install pdfbase

generate.ts

import PDFBase from 'pdfbase'

const client = new PDFBase('pk_live_...')

const pdf = await client.pdfs.create({

html: '<h1>Invoice #1042</h1>',

compress: 'medium',

watermark: { text: 'PAID' },

output: 'url'

})

// pdf.id → 'pdf_x7Kf9m'

// pdf.url → signed URL, 24h expiry

// pdf.pages → 3

// pdf.bytes → 48201

200 OK 184ms · 3 pages · 47KB

2.4M+

PDFs generated

184ms

Avg response time

99.97%

Uptime last 90 days

850+

Developer teams

“We replaced Puppeteer, pdf-lib, and a custom queue system with one PDFBase call. Saved us two sprints of infrastructure work.”

Marcus Reynolds

Lead Engineer, Docflow

“The debug mode alone is worth it. Being able to see exactly why a PDF render failed — screenshot, console logs, failed assets — is a game changer.”

Sarah Kim

CTO, InvoiceStack

“We hooked PDFBase into our AI agent via MCP and it just worked. Our agent generates client reports autonomously now. Zero prompt engineering needed.”

James Thornton

Founder, AgentOps

Five libraries. 400MB of Docker bloat.

Before

$ npm install puppeteer

$ npm install pdf-lib

$ npm install tesseract.js

$ npm install @gotenberg/client

$ npm install handlebars

dependencies

400MB

Chromium binary

20h+

to production

APIs to learn

After

$ npm install pdfbase

That's it. That's the stack.

Zero infrastructure to manage

Full Chromium rendering — no CSS hacks

One SDK, every operation

30 seconds to first PDF

$0.01

per PDF. Failed requests are free.

Replace your stack in 30 seconds

Every operation. One endpoint each.

POST /v1/pdfs

Generate

Send HTML or a URL, get a pixel-perfect PDF back. Full Chromium rendering — CSS Grid, Flexbox, web fonts, JavaScript. Everything works exactly like your browser.

wait_until: networkidle debug: true compress: medium

POST /v1/templates/:id/render

Templates

Save HTML + Handlebars once. Render with JSON data. Invoices, contracts, reports — at any scale.

{"customer": "Acme",
"items": [...]}

POST /v1/pdfs/merge

Merge & Split

Combine multiple PDFs into one. Split by page ranges. Reorder pages.

POST /v1/pdfs/watermark

Watermark

Stamp text or images on existing PDFs. Configurable opacity, position, rotation.

CONFIDENTIAL

POST /v1/extract

Extract + OCR

Pull text, tables, and structured data. OCR for scanned documents. Feed directly to your LLM.

PDF

JSON

POST /v1/convert

Office to PDF

Send DOCX, XLSX, PPTX. Get pixel-perfect PDFs. No LibreOffice to deploy.

.docx

.xlsx

.pptx

POST /v1/batch

Batch Processing

Queue thousands of PDFs in one call. Webhook when done. Idempotency keys prevent duplicates.

bat_8fK2mA · 2,400 / 2,400 done

POST /v1/pdfs/thumbnail

Thumbnails

Generate PNG previews of any PDF page. Perfect for file browsers, search results, dashboards.

Debug mode

Screenshot + console log + failed resources on every generation

Chromium pinned

Exact version documented. No rendering drift between deployments

Request replay

Replay any request from the dashboard. Debug production without code

Questions before you commit.

Is my data secure?

All API traffic is encrypted via TLS 1.3. Your PDFs are stored on encrypted volumes with signed URLs that expire in 24 hours. We never access, read, or train on your documents. You can also skip storage entirely and receive the PDF bytes directly in the response.

How reliable is the API?

We run on dedicated containers with warm Chromium instances — no cold starts. Average response time is under 200ms for simple documents. Status page is public at status.pdfbase.dev with real-time uptime tracking.

What happens when I run out of credits?

API calls return a clear 402 error — nothing breaks silently. You can set up auto-refill to automatically buy your preferred credit pack when you hit a threshold. Dashboard shows real-time usage so you’re never surprised.

Can I migrate away if I need to?

No lock-in. Your templates are plain HTML + Handlebars files you own. The API follows standard REST patterns — swapping to another provider means changing one HTTP endpoint. We don’t want you to stay because you’re trapped. We want you to stay because it’s better.

Why not just self-host Puppeteer?

You absolutely can — and for one template under 100 PDFs/month, you probably should. But once you need browser pooling, queue management, retry logic, storage, webhooks, and rendering across Chromium versions that don’t randomly break your CSS, the infrastructure tax adds up fast. That’s what we handle.

Do failed requests cost credits?

No. If a request fails for any reason — bad HTML, timeout, server error — you pay nothing. Only successful PDF generations deduct credits.

Technical Questions

What is PDFBase?

PDFBase is a REST API that handles every PDF operation in a single platform. Send HTML or a URL and get a pixel-perfect PDF back, powered by full Chromium rendering. Beyond generation, PDFBase supports merging, splitting, watermarking, compressing, text extraction, OCR, AI-powered structured data extraction, Office-to-PDF conversion (DOCX, XLSX, PPTX), thumbnail generation, and batch processing with webhooks. It comes with TypeScript and Python SDKs, an MCP server for AI agents, and an OpenAPI 3.1 spec. No Chromium to manage, no Docker to configure — just API calls.

How does PDFBase compare to Puppeteer for PDF generation?

Puppeteer is a great open-source tool for browser automation, and you can use it to generate PDFs. However, self-hosting Puppeteer means managing Chromium binaries (~400MB), handling browser pooling, queue management, retry logic, and storage yourself. PDFBase uses Chromium under the hood but handles all that infrastructure for you. You get the same rendering quality with a simple API call, plus additional features like watermarking, merging, OCR, and debug mode. For small-scale projects under 100 PDFs/month, Puppeteer works fine. For production workloads, PDFBase eliminates the operational overhead.

What file formats does PDFBase support?

PDFBase supports a wide range of input and output formats. For PDF generation, you can send raw HTML, a live URL, or Handlebars templates with JSON data. For conversion, PDFBase accepts Microsoft Office formats including DOCX (Word), XLSX (Excel), and PPTX (PowerPoint). For extraction, you can send any PDF and get back plain text, structured tables, or AI-extracted data as JSON. PDFBase also generates PNG thumbnail images from PDF pages.

Does PDFBase have OCR capabilities?

Yes. PDFBase includes built-in OCR for scanned documents and image-based PDFs. The extraction endpoint can pull text from scanned pages and return it as structured data. PDFBase also offers AI-powered extraction that goes beyond basic OCR — it can understand tables, invoices, receipts, and forms, returning structured JSON that you can feed directly to your application or LLM pipeline.

How much does PDFBase cost?

PDFBase uses a prepaid credit model with no subscriptions or overage fees. Every account starts with 100 free credits (no credit card required). Paid plans are: Starter at $10 for 1,000 credits ($0.01/credit), Growth at $50 for 5,500 credits ($0.009/credit, 10% savings), and Scale at $200 for 25,000 credits ($0.008/credit, 25% savings). Credits are valid for 12 months. Failed requests are free — only successful generations deduct credits.

What programming languages does PDFBase support?

PDFBase provides official SDKs for TypeScript/JavaScript (npm install pdfbase) and Python (pip install pdfbase). Since PDFBase is a standard REST API with JSON request/response bodies, you can use it from any programming language that can make HTTP requests — Go, Ruby, PHP, Java, C#, Rust, or any other language. The API follows OpenAPI 3.1 specification, so you can also auto-generate client libraries for your preferred language.

Can AI agents use PDFBase?

Yes, PDFBase is built for AI agents from day one. It includes an MCP (Model Context Protocol) server that works with Claude and other AI assistants, allowing agents to discover and call PDF tools automatically. The API is designed for machine consumption: JSON in, URL out, structured error responses, idempotent endpoints, and retry-safe operations. The full OpenAPI 3.1 spec means any agent framework can integrate PDFBase as a tool with zero manual configuration.

How does PDFBase compare to DocRaptor, Prince, and wkhtmltopdf?

Each tool has its strengths. wkhtmltopdf is free and open-source but uses an outdated WebKit engine that struggles with modern CSS (no Grid, limited Flexbox). Prince produces excellent print-quality PDFs and has superior CSS paged media support, but it requires a per-server license starting at $3,800. DocRaptor is a hosted API powered by Prince, offering great quality but at a higher price point with monthly subscriptions. PDFBase uses full Chromium rendering (the same engine as Chrome), so modern CSS just works. It is pay-per-use with no subscriptions, and includes features beyond generation — merge, split, watermark, OCR, extraction, and Office conversion — all in one API. If you need CSS paged media perfection, Prince is hard to beat. If you need a modern, full-featured PDF platform with simple pricing, PDFBase is a strong choice.