BlogLLM Benchmark Scores as Published, Side by Side
MMLongBench-Doc: Results as Published
Every result the catalog holds for MMLongBench-Doc: 13 models from 8 developers, read from 3 pages. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Multimodal UnderstandingMaintainer Yubo Ma and co-authors (Nanyang Technological University and others)Introduced 2024Data version 2026103001, as of 2026-10-30
MMLongBench-Doc tests question answering over long PDF documents that mix text, images, charts, tables, and layout. The current paper counts 1,062 expert-written questions on 130 documents, with an average of 49.4 pages and 20,971 text tokens. About a third of the questions need evidence from more than one page, and 22.8% are written so the document does not contain the answer. The reported score is an F1 score over these answers, including the unanswerable ones.
MMLongBench-Doc
Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.