SEA-HELM (Indonesian) - Metaphor Understanding: leaderboard

Metric: Normalized Score. Source: leaderboard.sea-lion.ai. 44 models tracked.

Top models

#ModelScore
1Gemma 4 31B (IT)83.33
2Qwen 3.6 27B78.63
3Qwen 3.5 122B A10B76.34
4Qwen 3.5 27B75.92
5Llama 3.3 70B Instruct75.7
6Gemma 4 26B A4B (IT)75.57
7Gemma 4 12B (IT)75.48
8Qwen 3 VL 32B Instruct74.98
9Llama 4 Scout Instruct72.42
10Mistral Medium 3.572.06
11Gemma 3 12B (IT)69.77
12Gemma 3 27B (IT)69.26
13Qwen 3 VL 8B Instruct68.33
14Qwen 3.6 35B A3B68.13
15gemma-4-E4B-it67.74

Interactive version: theaggregate.ai/benchmark?slug=sea-helm-indonesian-metaphor-understanding · How It Works · Data refreshed daily, snapshot 2026-09-19.