reasoning benchmark
MEGA MLQA
MLQA as part of the MEGA (Multilingual Evaluation of Generative AI) benchmark suite. A multi-way aligned extractive QA evaluation benchmark for cross-lingual question answering across 7 languages (English, Arabic, German, Spanish, Hindi, Vietnamese, and Simplified Chinese) with over 12K QA instances in English and 5K in each other language.
Updated Aug 7, 2026
Published models2
Registry coverage2
MetricScore
EvidenceB
MEGA MLQA leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | MI | 65.3% | 100.0% | 2 | C | |
| 2 | MI | 61.7% | 0.0% | 2 | C |
Score distribution
Top published rows on the benchmark's original scale.
MEGA MLQA highlights
The top published results on this benchmark's own scale.
What is MEGA MLQA?
Definition and scoring fields from the benchmark registry.
MLQA as part of the MEGA (Multilingual Evaluation of Generative AI) benchmark suite. A multi-way aligned extractive QA evaluation benchmark for cross-lingual question answering across 7 languages (English, Arabic, German, Spanish, Hindi, Vietnamese, and Simplified Chinese) with over 12K QA instances in English and 5K in each other language.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MEGA MLQA
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- mega-mlqa|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MEGA MLQA.
Which model scores highest on MEGA MLQA?
Phi-3.5-MoE-instruct is currently ranked first with 65.3%.
What does MEGA MLQA measure?
MLQA as part of the MEGA (Multilingual Evaluation of Generative AI) benchmark suite. A multi-way aligned extractive QA evaluation benchmark for cross-lingual question answering across 7 languages (English, Arabic, German, Spanish, Hindi, Vietnamese, and Simplified Chinese) with over 12K QA instances in English and 5K in each other language.
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.