Global-MMLU-Lite
Multilingual
text
+
+
+
+
About
Global-MMLU-Lite is a streamlined version of Global-MMLU containing 200 culture-specific and 200 culture-agnostic samples per language across 15 human languages. This lightweight benchmark provides efficient evaluation of AI models' multilingual and multicultural understanding while maintaining the core assessment of global knowledge diversity and cultural sensitivity with reduced computational requirements.
+
+
+
+
Evaluation Stats
Total Models14
Organizations1
Verified Results0
Self-Reported14
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
14 models
Top Score
89.2%
Average Score
69.6%
High Performers (80%+)
4Top Organizations
#1Google
14 models
69.6%
+
+
+
+
Leaderboard
14 models ranked by performance on Global-MMLU-Lite
| License | Links | ||||
|---|---|---|---|---|---|
| Jun 5, 2025 | Proprietary | 89.2% | |||
| May 20, 2025 | Proprietary | 88.6% | |||
| May 20, 2025 | Proprietary | 88.4% | |||
| Jun 17, 2025 | Creative Commons Attribution 4.0 License | 81.1% | |||
| Feb 5, 2025 | Proprietary | 78.2% | |||
| Mar 12, 2025 | Gemma | 75.1% | |||
| Mar 12, 2025 | Gemma | 69.5% | |||
| May 20, 2025 | Proprietary | 69.1% | |||
| May 20, 2025 | Gemma | 64.5% | |||
| Jun 26, 2025 | Proprietary | 64.5% |
Showing 1 to 10 of 14 models