Introducing ParseBench: The First Document Parsing Benchmark for AI Agents

LlamaIndex is open sourcing the first document OCR benchmark for the agentic era, ParseBench. Document parsing is the foundation of every AI agent that works with real-world files. But until now, there hasn't been a benchmark that measures parsing quality the way agents actually need it. ParseBench is an open-source benchmark of ~2,000 human-verified enterprise document pages with 167,000+ test rules across five dimensions that matter most: tables, charts, content faithfulness, semantic formatting, and visual grounding. LlamaParse Agentic was the only method competitive across all five dimen…