CC-OCR: Results as Published

CC-OCR tests reading and structured understanding of text in images. It has four tracks, multi-scene text reading, multilingual text reading, document parsing, and key information extraction, with 39 subsets and 7,058 fully annotated images, of which 41 percent are released here for the first time. Text reading is scored with an F1 over matched text units, document and formula parsing with normalized edit distance, tables with a tree-edit similarity, and key information extraction with a field-level F1. The paper does not define one metric that is used for every track.

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.