ExtractBench: schema-guided enterprise document extraction (LlamaIndex leaderboard)

Scores schema-guided extraction from 370 enterprise documents and 4,869 pages on value F1, record completeness, word and page grounding and cost, split by short, medium and long documents.

ModelsModels with a result.26
Top resultBest result on this benchmark.95.3%GPT-6.1 Sol
Top-3 spreadPoints from first to third.0.7 pts
YearYear of release.2026

Result and price

21 of 26 results

Results

26 results
#ModelResultThe score from the source.
03GPT-6 Sol94.6%
06GPT-5.593.3%
10GPT-6 Luna91.0%
13Qwen3.6-35B-A3BAbove fitFar above this model's index.88.1%
14Qwen3.5-35B-A3BAbove fitFar above this model's index.88.0%
18Kimi K383.2%