BlogLLM Benchmark Scores as Published, Side by Side
TyDi QA: Results as Published
Every result the catalog holds for TyDi QA: 4 models from 1 developer, read from 1 page. 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 Google Research (Jonathan H. Clark and co-authors)Introduced 2020Data version 2026103001, as of 2026-10-30
TyDi QA is a question-answering benchmark in 11 typologically diverse languages, with 204K question-answer pairs. People wrote the questions in each language because they wanted the answer and did not already know it, and the text was not translated from English. The primary tasks are to pick the answering passage, or to mark a minimal answer span, including yes, no, or no answer. The official score is an F1 computed per language and then averaged over the languages other than English.
TydiQA (F1)
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| Llama 4 Maverick (pre-trained)Meta, released 2025-04-05 | 31.7% | Meta, Meta Llama 4 model cardchecked 2026-10-15 | - |
| Llama 4 Scout (pre-trained)Meta, released 2025-04-05 | 31.5% | Meta, Meta Llama 4 model cardchecked 2026-10-15 | - |
| Llama 3.1 405B (pre-trained)Meta, released 2024-07-23 | 34.3% | Meta, Meta Llama 4 model cardchecked 2026-10-15 | - |
| Llama 3.1 70B (pre-trained)Meta, released 2024-07-23 | 29.9% | Meta, Meta Llama 4 model cardchecked 2026-10-15 | - |
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.