ExtractBench Pro: long-tail document extraction challenges (LlamaIndex)

A harder companion to ExtractBench: 100 synthetic documents and 2,622 pages across 86 document types that test document conventions, house vocabulary, counting rules, drawn relations and charts, scored by perfect-document rate.

ModelsModels with a result.7
Top resultBest result on this benchmark.—
Top-3 spreadPoints from first to third.—
YearYear of release.2026

Result and price

7 of 7 results

Results

7 results
#ModelResultThe score from the source.
—GPT-6 Astra58.0%
—GPT-6.1 Sol12.0%
—GPT-6 Luna0.0%