reasoning benchmark
MEGA TyDi QA
TyDi QA as part of the MEGA benchmark suite. A question answering dataset covering 11 typologically diverse languages (Arabic, Bengali, English, Finnish, Indonesian, Japanese, Korean, Russian, Swahili, Telugu, and Thai) with 204K question-answer pairs. Features realistic information-seeking questions written by people who want to know the answer but don't know it yet.
Updated Aug 7, 2026
Published models2
Registry coverage2
MetricScore
EvidenceB
MEGA TyDi QA leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | MI | 67.1% | 100.0% | 2 | C | |
| 2 | MI | 62.2% | 0.0% | 2 | C |
Score distribution
Top published rows on the benchmark's original scale.
MEGA TyDi QA highlights
The top published results on this benchmark's own scale.
What is MEGA TyDi QA?
Definition and scoring fields from the benchmark registry.
TyDi QA as part of the MEGA benchmark suite. A question answering dataset covering 11 typologically diverse languages (Arabic, Bengali, English, Finnish, Indonesian, Japanese, Korean, Russian, Swahili, Telugu, and Thai) with 204K question-answer pairs. Features realistic information-seeking questions written by people who want to know the answer but don't know it yet.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MEGA TyDi QA
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- mega-tydi-qa|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MEGA TyDi QA.
Which model scores highest on MEGA TyDi QA?
Phi-3.5-MoE-instruct is currently ranked first with 67.1%.
What does MEGA TyDi QA measure?
TyDi QA as part of the MEGA benchmark suite. A question answering dataset covering 11 typologically diverse languages (Arabic, Bengali, English, Finnish, Indonesian, Japanese, Korean, Russian, Swahili, Telugu, and Thai) with 204K question-answer pairs. Features realistic information-seeking questions written by people who want to know the answer but don't know it yet.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
2 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.