SDK benchmarks
How fast does the viewer render? How long does OCR take on a 25 MB file? What server do I need? What’s the real PDF/A pass rate? We tested it and wrote it down — methodology, raw numbers, and per-document breakdowns in every report.
Web Viewer SDK
First-page render times across 22 documents, browser-only (WASM) and server-rendered (Document Engine) modes
Document Engine
Server-side PDF processing component — per-operation latency and infrastructure sizing
PDF/A and PDF/UA
veraPDF-validated compliance rates across multiple vendors
Methodology and versions used are outlined in each report.
By submitting this form, you agree to Nutrient’s Privacy Policy and Terms of Service.
Web Viewer SDK runs in the browser; Document Engine runs on the server. Either works on its own, or you can pair them so Document Engine handles rendering and processing for Web Viewer SDK on large files. The reports cover all three setups: Web Viewer SDK standalone, Document Engine standalone, and the two combined.
These four reports cover Web Viewer SDK (client-side viewer), Document Engine (server-side PDF processing component), and PDF/A and PDF/UA conversion. Together they span the full path — from the browser to the server to compliance output.
Measures how fast your users see the first page. Tested on 22 real-world documents — invoices, CAD drawings, scanned archives, and more — comparing Web Viewer SDK standalone rendering against Web Viewer SDK powered by Document Engine, our server-side PDF processing component.
How long does each API call take? Document Engine is a Docker-based server component that handles OCR, text extraction, search, PDF export, conversion, redaction, and annotation operations. This report focuses on per-operation latency across simple, medium, and complex document profiles.
How much can one server handle? This report focuses on capacity and sizing, presenting a tested configuration (ARM64, 16 vCPU, 32 GiB) under a continuous concurrency ramp from 1 to 32, so you can map measured throughput to your expected load before provisioning.
Compares pass rates for archival (PDF/A) and accessible (PDF/UA) conversion, validated by veraPDF — the same tool auditors use. Nutrient is compared against up to seven vendors — including Apryse, PDF-Tools, Foxit, and IronPDF — on 3,157 files.
You get simpler infrastructure with Web Viewer SDK alone, or faster load times when paired with Document Engine. The benchmarks test both setups on the same 22 documents so you can see where the tradeoff sits for your documents, traffic, and infrastructure.
“What server do I need?” depends on your document sizes, operations, and concurrency. The reports give you a tested server configuration under real load — so your first deployment is based on measured throughput, not rough estimates.
Every vendor claims PDF/A and PDF/UA support. The question is what percentage actually passes validation. An 85 percent pass rate means 15 percent of your output fails — a problem you discover during an audit. The reports use veraPDF across 3,157 files and multiple vendors.
We test on real-world document types, not synthetic test files. The Web Viewer SDK benchmarks cover invoices, contracts, CAD drawings, floor plans, medical documents, legal filings, newspapers, magazines, and scanned archives — 22 documents across six categories. The compliance benchmarks use 3,157 files from diverse sources.
The web rendering report includes head-to-head data against Apryse WebViewer on all 22 documents. The PDF/A and PDF/UA reports compare Nutrient against multiple vendors, including Apryse, PDF-Tools, Foxit, and IronPDF. The infrastructure and server operation reports focus on Document Engine, Nutrient’s server component for PDF processing. They’re for sizing your deployment, not for competitive comparison.
Every report documents the full test setup: hardware, software versions, network conditions, document set, and statistical method. The web benchmarks run on 22 documents we didn’t choose for favorable results — they include document types where rendering is slow (large scans, complex layers). The compliance benchmarks use 3,157 files per vendor, with veraPDF as an independent validator. We publish methodology so you can reproduce the tests or challenge the setup.
Performance depends on your documents, hardware, network conditions, and how your application calls the SDK. These benchmarks are reference points, not guarantees. The test conditions are designed to reflect production reality — the web tests use a Wi-Fi-throttled network, and the server tests include the full HTTP layer with concurrent load.
The PDF/A report shows veraPDF conformance rates across seven vendors on 3,157 files. veraPDF is the same open source validator auditors and archival institutions use. The report also tracks critical failures (unrecoverable crashes during conversion) — if your pipeline needs to run unattended, that number matters as much as the pass rate.
Yes. The PDF/UA report tests 3,157 files across four vendors with veraPDF validation. It covers conformance rates, critical failures, and throughput. If you need to meet Section 508, EN 301 549, or other accessibility mandates, this report shows how each vendor’s output holds up at scale — not just whether the feature exists.
Yes. Our solutions engineers can test with your actual documents on your target infrastructure. This is the most reliable way to validate performance for your specific use case. Schedule a call to set this up.
Yes. They’re designed for internal evaluation. Share them with engineering, product, DevOps, compliance, and procurement — whoever is involved in the decision.
