BlogLLM Benchmark Scores as Published, Side by Side
WMT24++: Results as Published
Every result the catalog holds for WMT24++: 26 models from 10 developers, read from 5 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 MultilingualMaintainer Daniel Deutsch and co-authors (Google)Introduced 2025Data version 2026103001, as of 2026-10-30
WMT24++ is an English-to-many translation test. It reuses the WMT24 English sources, 998 paragraphs from literary, news, social, and speech text, and adds human-written references and post-edits so the set covers 55 languages and dialects. Systems translate those English sources, and the translations are scored with automatic metrics. The paper's main metric is MetricX-24, with and without a reference, and the appendix also reports BLEU, ChrF, XCOMET, and COMETKiwi, using the post-edits as references.
WMT24++ (XCOMET-XXL)
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.