Image & vision

ExtractBench (LlamaIndex)

by LlamaIndex

What is ExtractBench (LlamaIndex)? ExtractBench (LlamaIndex) is an image & vision eval made by LlamaIndex. Schema-guided extraction from enterprise documents: given a full document and a JSON schema, a system must return every requested value with evidence. 370 documents, 4,869 pages, 67 document types and 8 business domains, scored without an LLM judge on value F1, record completeness, word- and page-level grounding and measured cost. Built by LlamaIndex, which also sells extraction tools that appear on its leaderboard.

Page last updated 2026-10-18 · Evals Review

Editor note

Schema-guided extraction from enterprise documents: given a full document and a JSON schema, a system must return every requested value with evidence. 370 documents, 4,869 pages, 67 document types and 8 business domains, scored without an LLM judge on value F1, record completeness, word- and page-level grounding and measured cost. Built by LlamaIndex, which also sells extraction tools that appear on its leaderboard.

What it tests
Document-extraction systems, VLMs and coding agents filling a schema from long business documents.
Original source
www.extractbench.ai/
Last updated
Repo last commit 2026-10-20
License
Apache-2.0

Facts come only from the linked source (checked 15–16 Oct 2026). Credit belongs to the creators. We don't score or rank evals: how we review.

Reviews

No reviews yet.

Reviews are written by people who sign in with GitHub, one per account per eval. We can't confirm that each reviewer used the eval; we remove reviews that break our rules, never just for being negative, and we never add, buy or edit reviews. Averages, when shown, come only from these reviews.

Reviews from practitioners who sign in with GitHub open soon.

Related image & vision evals

  • MMMU — Expert-level multimodal understanding and reasoning.
  • MathVista — Mathematical reasoning over figures, plots and images.
  • MMBench — Broad perception and reasoning skills of vision-language models.
  • ChartQA — Question answering over charts.

All Image & vision evals · Comparison table