LlamaIndex Launches ExtractBench on Kaggle

The official LlamaIndex account announced that ExtractBench is now live on Kaggle. The benchmark evaluates schema-guided document extraction on files that commonly disrupt downstream agents and workflows. It focuses on long record lists, noisy scans, handwriting, and complex tables. The effort draws from 370 enterprise documents spanning 8 business domains and 67 document types. A light gray promotional graphic accompanied the post. The announcement presents the benchmark as a public resource for testing extraction performance on challenging real-world materials.