Is this an OCR benchmark?

It is an invoice extraction benchmark kit for field and line-item extraction, not a generic OCR benchmark.

Can I run it locally?

Yes. The evaluator and supporting files ship inside the pack.

Is this real invoice data?

No. The invoices are synthetic.

What do I receive after purchase?

You receive a download link to the benchmark kit ZIP, typically within 24 hours of payment. The pack includes invoice PDFs, labels, evaluator, sample reports, benchmark protocol, and supporting docs. Everything runs locally.

Invoice extraction benchmark kit

Invoice extraction benchmarking, packaged.

Compare invoice extraction approaches with a packaged benchmarking kit that includes ground truth, evaluator, reports, and a clear protocol. Run it locally.

Download free sample View pricing

What it is

packaged invoice extraction benchmark kits for US English invoices
built for local evaluation, regression checks, and internal demos
field and line-item extraction benchmark

Inside the benchmark kit

Representative screenshots from the benchmark kit.

Representative minimal invoice screenshot from the benchmark pack — **Representative clean invoice** Minimal layout from the benchmark kit.

Representative dense invoice screenshot from the benchmark pack — **Structured layout example** Denser table-oriented layout from the kit.

Representative light-noisy invoice screenshot from the benchmark pack — **Light-noisy example** Simulated scan noise applied to the same layout.

Sample benchmark report summary preview — **Sample benchmark report** Core metrics and coverage summary from the evaluator output.

Benchmark card overview preview — **Benchmark card overview** Compact summary of scope, coverage, and limitations.

How it works

Three steps, local workflow.

1. Run your extractor

Use the included invoice PDFs with your model, OCR pipeline, or parser.

2. Score predictions

Run the local evaluator (standalone Python script, no dependencies) against the included labels and inspect the report.

3. Compare changes

Reuse the same benchmark when you update prompts, models, or parsing logic.

You can build an invoice benchmark yourself. Most teams spend days on document generation, label alignment, and evaluator edge cases before running a single comparison. This benchmarking kit ships ready to score.

Pricing

Three tiers. One workflow.

Free Sample

Inspect the kit and evaluator workflow. Direct download.

Starter

$299

The main paid benchmark for comparisons, demos, and regression checks.

Full

$990

Broader coverage for repeated evaluation and higher-confidence comparison work.

Download free sample View pricing details

Limitations

What this is not for.

not production invoice processing
not compliance validation
not a hosted benchmark service
not a replacement for your private documents

Next step

Try the free sample first.

Start with the free sample. Move to Starter when you want the main paid benchmark for real evaluation work.

Download free sample View pricing

Need a custom benchmark or dataset? Get in touch — we build benchmarks and datasets for any document type or domain.