10 Best HTML to PDF APIs for Developers in 2026

Running Puppeteer for a few invoices is different from maintaining it under concurrent load. JavaScript timeouts, memory use, cold starts, and font differences between development and production can make self-hosted rendering harder to operate as demand grows.

Moving to a hosted API swaps those problems for two questions, which engine renders your HTML and what counts as one billable render. 

Those two questions drive the decision more than the headline plan price, because the billing models across these APIs have almost nothing in common. PDFShift charges 1 credit per 5 MB of output, PDFCrowd 1 credit per 0.5 MB, PDF.co 9 credits per page, Browserless one unit per 30 seconds of browser time, and DocRaptor one document whatever its size or page count.

We compare our Foxit PDF Services API with 9 alternatives: DocRaptor, PDFShift, ConvertAPI, PDFCrowd, Api2Pdf, Browserless, PDF.co, Nutrient DWS Processor API, and Gotenberg. 

The comparison covers rendering engines, JavaScript handling, layout controls, output options, pricing, free allowances, limits, SDKs, and hosting, using public documentation, API specifications, pricing pages, and GitHub repositories.

Top picks by use case

The best HTML to PDF API depends on where your layout logic lives and how your renders will be billed. 

For products that render HTML and then process the resulting PDF, the Foxit PDF Services API covers both steps on one credit pool. 

For print-styled documents with complex pagination, DocRaptor’s Prince engine applies CSS paged media directly. The table below maps each pick to the use case it fits best.

These picks highlight different use cases, not exclusive capabilities. Check each API’s features and limits against your full workflow.

Create a free Foxit developer account and convert your first HTML file or URL to PDF with the PDF Services API.

How we compared these HTML to PDF APIs

Every fact in this article comes from public documentation, API specs, pricing pages, and GitHub, all checked on 16 Oct 2026. Recheck any price before making a purchasing decision, since billing details change.

Compare inputs and engines, JavaScript timing, layout controls, output options, pricing, free allowances, limits, SDKs, and hosting. Where a vendor does not publish a number or engine name, the table notes that gap as of 16 Oct 2026.

Three APIs in this list do not name their HTML rendering engine in their public documentation, namely Foxit PDF Services API, PDFCrowd, and PDF.co. ConvertAPI names Chrome headless. Nutrient’s overview names Chromium, but the generation guide does not and conflicts with the overview on JavaScript support.

We found no independent, reproducible fidelity or speed benchmark of hosted HTML-to-PDF APIs in the sources reviewed. The comparisons that rank for this query are vendor-written.

All prices are in USD unless otherwise noted. Billing period is named beside every price.

The 10 HTML to PDF APIs in detail

Start with the engine and the controls your template needs. Chromium-based APIs such as PDFShift, Browserless, Gotenberg, and Api2Pdf’s Chrome route use a browser print engine. DocRaptor uses Prince for CSS paged media. Check the engine comparison table for differences in JavaScript, headers, and footers.

Foxit PDF Services API

Our PDF Services API is a cloud REST API for automating PDF workflows. 

After converting HTML to PDF, you can compress, merge, watermark, protect, or OCR the result, or send it for signature with our eSign API, using the same developer account and shared credit pool.

Our pricing calculator and credit reference list Business at $4,500/year for 15,000 credits, while the pricing page’s plan cards show a custom Volume tier. Confirm the available plan with our team before purchasing.

Where Foxit PDF Services API stands out

A two-page invoice and a 150-page annual report each use 1 conversion credit within the supported limits. That keeps page count out of the conversion cost calculation, though follow-on operations consume additional credits.

The same credit pool covers the steps that typically follow a render. Compress the file for email delivery, merge it with a cover page or terms document, watermark it, protect it with a password, auto-tag it for accessibility, or send it for signature with eSign, all without touching a second vendor or a second billing system.

For agentic workflows, the Foxit MCP Server exposes pdf_from_html and pdf_from_url as tools that LLM agents can call directly, without custom REST client code.

The Foxit PDF Services page confirming MCP server access comes at no added cost beyond existing API credentials.

What to consider

The API reference does not document margin, header and footer, or wait-for-selector parameters. Layout control happens inside the HTML itself, using @page rules and print CSS. If your template relies on request-level margin or footer injection, you’ll need to build those into the markup before upload.

For uploaded HTML, conversion follows four steps: upload the file, submit the job, poll for completion, and download the PDF. Polling and downloads consume no credits. Build your error handling around task status rather than a single conversion response. Our htmlToPDF.py sample shows this flow in Python, with a Node.js equivalent alongside it.

The htmlToPDF.py sample on GitHub, showing the working Python implementation of the async flow described above.

The four-step async flow. Only the submit step consumes a credit; polling and downloading are free.

Credits are sold as an annual pool with no rollover. If your rendering volume is spiky or hard to predict, estimate conservatively and track consumption through the developer console.

The Foxit developer console’s dashboard, showing the Credits Used tile alongside the active plan’s included credit balance.

Best for

Products where the HTML render is the first step of a longer document workflow, such as generating an invoice and then compressing, watermarking, and sending it for signature on the same account.

Getting started

Create a free Foxit developer account and convert your first HTML file or URL to PDF with the PDF Services API.

The Foxit developer sign-up page, confirming no credit card is required to get started.

DocRaptor

DocRaptor is built on Prince, a CSS typesetting engine that treats HTML as print layout input rather than a browser page. 

It’s the go-to for documents whose pagination must match a designed layout, including long reports, financial statements, and books.

DocRaptor’s full tier ladder, from the free plan through Silver’s 40,000 documents a month.

Where DocRaptor stands out

Prince applies CSS paged media to page size, running headers and footers, margins, and page numbering. Pagination is designed in CSS rather than patched with browser workarounds, which makes complex multi-page layouts consistent and predictable.

DocRaptor’s per-document billing is indifferent to file size or page count. A 200-page regulatory report costs the same credit as a two-page invoice, so teams with large, complex documents get more predictable costs than they would with size-based billing.

PDF/A, PDF/UA, and PDF/X profiles are available at render time with no extra API call. If your compliance requirements include archival or accessibility profiles, they’re applied in the same request that generates the document.

What to consider

JavaScript is off by default. DocRaptor offers its own JavaScript engine and Prince’s separate engine, enabled through different parameters. Test client-rendered charts, dynamic tables, and single-page-app templates with the engine you choose; Chrome previews alone do not establish compatibility.

Rendering runs only in AWS us-east-1. If data sovereignty requires keeping rendering in the EU or Asia-Pacific region, DocRaptor has no regional routing option.

Async jobs send callbacks only on success. A failed async job produces no callback, so your polling logic must handle both successful and failed states.

Best for

Long, print-styled reports and documents whose pagination must match a designed print layout.

Getting started

Start with DocRaptor’s unlimited watermarked test documents on any plan to validate your template before buying a paid tier.

DocRaptor’s documentation on the test parameter, which produces unlimited watermarked documents that do not count against the plan limit.

PDFShift

PDFShift is a Chromium-based rendering API with plans for a range of volumes and version pinning that lets you choose when to change the Chromium version used for your templates.

PDFShift’s pricing calculator, confirming the Business tier at 25,000 credits for $99 a month. The page picks a single recommended tier from a credit-count slider rather than listing all tiers at once.

Where PDFShift stands out

Version pinning lets you choose Chromium 116 or 142 with the X-Processor-Version header rather than follow PDFShift’s default engine. It reduces one source of rendering changes, but you still need to test templates when their fonts, assets, or application code change.

The tier ladder runs from $9 a month to $999, covering a wide volume range without requiring a sales conversation. Direct export to your own S3, Google Cloud Storage, or R2 bucket means rendered PDFs land in your storage without an extra download step.

Size-based billing rewards lean templates. PDFs that are mostly text and vector graphics use fewer credits than image-heavy designs, giving you an incentive to optimize output size.

What to consider

Header and footer templates do not inherit your page CSS. Styles applied in the body are invisible inside the header or footer, which is a standard Chromium limitation. You need to duplicate any relevant styles inside those templates.

Image-heavy PDFs consume credits faster than plain ones, because 1 credit covers 5 MB of output. A rendering pipeline for photo-rich marketing materials will cost more than one for text-heavy invoices.

PDF/A output is not in the API schema as of 16 Oct 2026. If you need archival profiles, this API won’t meet that requirement.

Best for

High-volume rendering of modern HTML and CSS where cost per render and predictable engine behavior are the priorities.

Getting started

Use PDFShift’s 50 free monthly credits to test your templates and measure per-file credit consumption before choosing a paid tier.

ConvertAPI

ConvertAPI is a Chromium-based conversion API with CSS-selector page break controls and regional endpoints in the UK, EU, and US.

ConvertAPI’s annual-billing tier prices, which match the $30/$85/$160 figures cited in the feature table.

ConvertAPI’s HTML to PDF API reference page, naming Chrome headless as the engine, with a link to the OpenAPI schema beside it.

Where ConvertAPI stands out

The AvoidBreakElements, BreakBeforeElements, and BreakAfterElements parameters let you control page breaks by CSS selector in the API request. That’s a level of pagination control you’d normally need to patch into the HTML itself, and it’s available across all plans.

Regional endpoints in London, Amsterdam, Frankfurt, and two US regions let you choose where rendering runs. If you have EU data residency requirements, confirm the applicable storage, transfer, and contractual terms as well as the endpoint location.

A configurable timeout of up to 1,200 seconds gives slow pages more time to finish rendering. Check it against your application’s own request timeout and response-time requirements.

What to consider

The default media type is screen. If your stylesheet uses @media print rules, you need to set CssMediaType explicitly in each request. Missing this is easy and produces output that ignores your print styles entirely.

Each request can count as one or more conversions, and successful responses report ConversionCost. Failed requests are also billed. Review the ConversionCost field in your response to avoid surprises.

ConvertAPI’s compliance wording varies across its own pages. ISO 27001 and GDPR appear consistently; treat other compliance claims as requiring verification.

Best for

Invoices and reports that need explicit control over where pages break, processed in the UK, EU, or US.

Getting started

The 250 free conversions at sign-up are enough to validate page break behavior and media type handling on your template before purchasing a plan.

PDFCrowd

PDFCrowd offers the widest PDF/A coverage in this comparison, with variants from PDF/A-2a through PDF/A-4f, tagged PDF, and credit-based billing that scales with output size.

PDFCrowd’s credit-based tiers, priced at 1 credit per 0.5 MB of output.

Where PDFCrowd stands out

PDFCrowd supports PDF/A variants from PDF/A-2a through PDF/A-4f, including PDF/A-4e for engineering documents. Tagged PDF output is also available, but tagging alone does not establish accessibility compliance.

Occasional overage is not billed on PDFCrowd’s documented policy, and entry-tier pricing starts at $1 a month billed annually, which makes low-volume archival use affordable.

Zipped HTML upload with assets up to 300 MB handles complex templates with many local resources without requiring a public URL or a content delivery network.

What to consider

Entry plans cap JavaScript waits at 1 to 2 seconds and run one conversion at a time. Concurrency and wait limits scale with plan tier, so a low-cost entry plan may not handle templates that depend on slow client-side rendering.

At 0.5 MB per credit, image-heavy or font-rich PDFs consume more credits than smaller exports. Compare the cost for your actual output sizes, not credit counts alone, because vendors price credits differently.

Sustained overage beyond occasional instances can lead to account suspension according to PDFCrowd’s overage documentation. Monitor consumption carefully on smaller plans.

Best for

Archival or accessible PDFs from HTML at steady monthly volume, especially when choosing among PDF/A variants is a requirement.

Getting started

The 100-credit trial covers a month of testing, enough to measure real output sizes and credit consumption before choosing a tier.

Api2Pdf

Api2Pdf is a metered API that bills for actual compute time and bandwidth consumed, with per-call choice of engine, covering Headless Chrome, wkhtmltopdf, or LibreOffice.

Api2Pdf’s pay-as-you-go pricing page, including its own cost estimator for bandwidth and compute time.

Where Api2Pdf stands out

Api2Pdf estimates about $5.91 for its example of 10,000 small PDFs a month. For an unpredictable workload, compare that metered model with a subscription using your own render times and output sizes.

Each call chooses its engine independently. A batch that mixes HTML invoices (Chrome), legacy HTML files (wkhtmltopdf), and Office documents (LibreOffice) runs through the same API key without switching services.

The $1 monthly base fee after the first month is low enough that Api2Pdf is viable even for low-volume production use, such as a small internal tool that generates a handful of PDFs a week.

What to consider

Cost follows render time and output size, so a slow page with heavy JavaScript costs more than a fast static template. Profile your templates during the free month to understand what they’ll actually cost.

The wkhtmltopdf route runs on an archived, unmaintained engine (see “What happened to wkhtmltopdf” below). For new projects, the Chrome route is the maintained option.

The 90-second per-request cap is a hard ceiling. Templates that require long JavaScript execution or large resource loads need to fit within that window. V2 XL lifts it to 5 minutes at 10 times the compute cost.

Best for

Spiky or cost-sensitive workloads where paying for actual compute beats a fixed monthly tier.

Getting started

The one-month free period is the right time to run your real templates and compute your expected monthly cost before committing.

Browserless

Browserless runs Chrome as a managed service, exposing the same Puppeteer and Playwright controls your existing render code already uses, without you operating the browsers.

Browserless’s unit-based tiers, where one unit covers up to 30 seconds of browser time.

Where Browserless stands out

Existing Puppeteer render code moves over with minimal changes. The /pdf REST API accepts the same options as Puppeteer’s page.pdf(), so teams moving off self-hosted Chrome don’t need to rewrite their rendering logic.

The full session API supports Playwright with Chrome, WebKit, and Firefox for browser automation. Check PDF-export support separately rather than assuming that every supported browser can generate PDFs.

Three geographic regions (San Francisco, London, Amsterdam) give you control over where browser sessions run, reducing latency for geographically distributed workloads.

What to consider

Browserless charges browser time in blocks of up to 30 seconds per connection. Longer waits can increase usage when they cross a block boundary, so measure the full connection duration rather than assuming cost rises in direct proportion to render time.

PDFs that fail to render in a single request may need chunking by page range and merging, which adds request complexity. The /pdf endpoint cannot produce a single continuous long-page PDF.

Chrome’s tagged PDF output is available but Browserless explicitly notes it is not certified PDF/UA. If formal accessibility compliance is a requirement, verify this limitation before building on it.

Best for

Teams with existing Puppeteer or Playwright render code who want to stop operating the browsers themselves.

Getting started

The 1,000 free units a month (2 concurrent browsers) cover lightweight testing. Run your actual templates against the free tier to measure unit consumption before selecting a paid plan.

PDF.co

PDF.co is owned by Artifex Software (the same parent as DocRaptor) and places HTML to PDF alongside parsing, forms, barcode, and extraction endpoints on one shared credit balance, with Make and Zapier modules for no-code integration.

PDF.co’s annual-billing tiers, matching the 16,500/37,000/80,500 credit figures cited above.

Where PDF.co stands out

HTML to PDF shares a credit balance with parsing, form filling, barcode generation, compression, and splitting. An automation builder in Make or Zapier can chain all of these in one workflow without switching services or managing multiple API keys.

Per-page credits (9 per page) make PDF.co’s cost predictable for fixed-length documents. An invoice that’s always two pages always costs 18 credits, regardless of file size.

The 10,000-credit free trial gives you room to run a realistic volume of test documents and understand exactly what your credit consumption looks like before choosing a plan.

What to consider

At 9 credits per page, a 50-page report uses 450 credits and a two-page invoice uses 18. Personal’s 37,000 monthly credits would cover at most 82 reports or 2,055 two-page invoices if used only for rendering. Polling and other operations reduce those totals.

JavaScript always runs on load and cannot be disabled. If you’re rendering static HTML, there’s no way to skip the JavaScript execution time.

Synchronous calls stop at 30 seconds. Templates with heavy JavaScript or many external resources need the async route with webhook handling.

Best for

Automation builders on Make or Zapier who need HTML to PDF alongside other PDF jobs on one credit balance.

Getting started

The free 10,000-credit trial covers about 1,111 rendered pages, enough for realistic testing of your document types.

Nutrient DWS Processor API

Nutrient DWS Processor API converts HTML to PDF at 0.5 credits per conversion and turns HTML form fields into interactive, fillable PDF form fields, with the option to chain multiple operations in a single /build call.

Nutrient’s Processor API pricing, shown at the annual-billing rate (about 10% below the monthly prices cited in the feature table).

The Processor API overview page’s FAQ, naming Chromium as the engine and listing JavaScript among the web standards it supports, the first half of the conflict described below.

The PDF generation guide’s own callout, stating plainly that JavaScript assets are not supported, the second half of the conflict.

Where Nutrient DWS Processor API stands out

HTML form fields in your source markup turn into fillable PDF form fields in the output, with no post-processing required. If you’re generating application forms, order forms, or any document the recipient needs to complete, this capability is available directly in the conversion step.

Chaining operations in a single /build request reduces round trips. Convert HTML to PDF and apply PDF/A or PDF/UA in one call, rather than two separate API requests.

At 0.5 credit per HTML conversion, 1,000 credits cover 2,000 conversions if used only for that operation. Include any additional operations when estimating the cost of the full workflow.

What to consider

The overview and the generation guide conflict on JavaScript support. The generation guide states JavaScript assets are not supported; the overview says they are. Ship pre-rendered HTML until Nutrient clarifies this at the route level.

Free-plan output is watermarked unless you enable pay-as-you-go with a $25 minimum wallet load. Budget for that activation cost if you’re testing without a paid plan.

Best for

Static HTML templates such as applications or order forms that must become fillable PDFs.

Getting started

Review Nutrient’s HTML-to-PDF guide, then test your form-field conversion on the free tier (50 credits a month, watermarked) before selecting a plan.

Gotenberg

Gotenberg is an MIT-licensed Docker API wrapping Chromium and LibreOffice, with PDF/A, PDF/UA, and e-invoice output built in. 

Version 8.38.1 was released on 15 Oct 2026. There are no per-render license fees or vendor quotas, but you pay for hosting and take responsibility for operation and maintenance.

Gotenberg’s GitHub repository, confirming the MIT license and its Chromium/LibreOffice/wkhtmltopdf topic tags.

Gotenberg’s Releases page, confirming v8.38.1 as the current release.

Where Gotenberg stands out

Self-hosting Gotenberg lets you process HTML and PDFs on your own infrastructure. For data residency or air-gapped deployments, also control outbound requests, external assets, and webhook destinations; self-hosting alone does not prevent data from leaving the environment.

PDF/A, PDF/UA, merge, split, encrypt, and e-invoice formatting are all available in the same container, with no per-operation billing. Volume is limited only by the compute you allocate.

What to consider

You own the operational burden, including container updates, Chromium upgrades, concurrency tuning, and uptime. The 30-second default timeout is configurable, but scaling to high concurrency requires running multiple container instances and a load balancer in front of them.

header.html and footer.html cannot load external resources or run JavaScript, which is the same architectural constraint as other Chromium-based APIs.

Best for

Teams with data residency requirements, air-gapped environments, or high enough volume that self-hosting costs less than per-render fees.

Getting started

Pull the official Docker image, run it, and POST your HTML to the Chromium route. The MIT license means no activation step. Grab a ready-to-use sample index.html here and save it in the same directory you run the curl command from, so the –form ‘[email protected]’ line below has something real to upload.

Run this command from your terminal: