SEA-HELM: leaderboard
Holistic evaluation of LLMs across Southeast Asian languages covering NLP tasks, linguistics, cultural knowledge, and safety across Filipino, Indonesian, Thai, Vietnamese, and more.
Metric: Mean Score (%). Source: leaderboard.sea-lion.ai. Saturation forecast: Around January 2027. 58 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B | 76.17 | #200 |
| 2 | Gemma 4 26B | 74.81 | #292 |
| 3 | Qwen 3.5 122B A10B | 74.66 | #173 |
| 4 | Qwen 3.6 27B | 74.64 | #170 |
| 5 | Qwen 3.5 27B | 74.17 | #187 |
| 6 | Qwen 3.6 35B A3B | 73.86 | #226 |
| 7 | Qwen 3.5 35B A3B | 73.55 | #256 |
| 8 | Muse Glimmer 30B | 72.13 | #178 |
| 9 | Gemma 4 12B | 69.31 | #535 |
| 10 | Qwen 3.5 9B | 69.13 | #379 |
| 11 | Qwen 3 32B | 67.74 | #443 |
| 12 | Gemma 4 E4B | 66.3 | #767 |
| 13 | GPT-OSS-120B | 66.14 | #341 |
| 14 | Qwen 3 14B | 64.75 | #537 |
| 15 | Mistral Medium 3.5 | 64.71 | #353 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sea-helm · How It Works · Data refreshed daily, snapshot 2026-10-20.