SEA-HELM (Malay) - Safety: leaderboard
SEA-HELM's safety competency for Malay: combines its toxicity detection and SEA-Safeguard tasks. Normalized 0-100 score from AI Singapore's SEA-HELM leaderboard: (score - random baseline) / (maximum - baseline), floored at 0 and averaged over eight runs; higher is better.
Metric: Normalized Score. Source: leaderboard.sea-lion.ai. Saturation forecast: Around March 2028. 58 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 27B (IT) | 45.78 | #528 |
| 2 | Qwen 3.5 122B A10B (Reasoning) | 43.58 | #184 (Qwen 3.5 122B A10B) |
| 3 | Qwen 3 32B (Thinking) | 42.82 | #449 (Qwen 3 32B) |
| 4 | Llama-SEA-LION-v3-70B-IT | 42.69 | #407 |
| 5 | Qwen 3 VL 32B Instruct | 42.47 | #290 |
| 6 | Qwen 3.5 35B A3B (Reasoning) | 42.36 | #268 (Qwen 3.5 35B A3B) |
| 7 | MERaLiON-2-10B | 42.24 | #600 |
| 8 | Gemma 3 12B (IT) | 42 | #672 |
| 9 | Gemma 4 26B A4B (IT) (Reasoning) | 41.37 | #293 (Gemma 4 26B A4B (IT)) |
| 10 | Llama-SEA-LION-v3-8B-IT | 41.15 | #800 |
| 11 | Qwen 3.5 9B (Reasoning) | 41.03 | #391 (Qwen 3.5 9B) |
| 12 | Qwen 3.5 27B (Reasoning) | 40.8 | #192 (Qwen 3.5 27B) |
| 13 | Qwen 3.5 4B (Reasoning) | 40.76 | #514 (Qwen 3.5 4B) |
| 14 | Llama 3.3 70B Instruct | 40.72 | #541 |
| 15 | Gemma 3 4B (IT) | 39.91 | #997 |
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-malay-safety · How It Works · Data refreshed daily, snapshot 2026-11-05.