Benchmarking
Two ways to measure PDF workflows
Explore how document parsing performs on its own, and how that output affects the quality of information extracted by an LLM. ParseBench measures parser quality and performance; ExtractBench measures the accuracy of the final answers.
Extraction quality
ExtractBench
Compare how well an LLM extracts content from direct PDF file inputs and parser-generated Markdown, including tables, values, and document details.
View ExtractBenchParser performanceParseBench
Compare document parsers across quality, speed, installation footprint, and cost on a broad benchmark of document types.
View ParseBench