Llama-SEA-LION-v3-8B-IT: benchmark results
Provider: AI Singapore. Released 2024-12-18. Access: Open.
Aggregate Rating 1470 ± 27, rank #2946 of 10524 rated models, from 144 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SEA-HELM (Vietnamese) - Sentiment Analysis | 63.88 | Normalized Score | 93.2 |
| SEA-HELM (Indonesian) - SEA-Safeguard | 73.79 | Normalized Score | 87.8 |
| SEA-HELM (Malay) - Toxicity Detection | 14.36 | Normalized Score | 87.8 |
| SEA-HELM (Thai) - Summarization | 23.93 | Normalized Score | 82.4 |
| SEA-HELM (Indonesian) - Summarization | 17.15 | Normalized Score | 78.4 |
| SEA-HELM (Malay) - Safety | 41.15 | Normalized Score | 68.9 |
| ThaiSafetyBench - Malicious Uses | 10.29 | Attack Success Rate (%) | 65.2 |
| SEA-HELM (Tamil) - Summarization | 11.46 | Normalized Score | 63.5 |
| SEA-HELM (Vietnamese) - Summarization | 16.3 | Normalized Score | 62.2 |
| ThaiSafetyBench - Discrimination, Exclusion, Toxicity, Hateful, Offensive | 8.57 | Attack Success Rate (%) | 60.9 |
| SEA-HELM (Filipino) - SEA-Safeguard | 67.87 | Normalized Score | 58.1 |
| ThaiSafetyBench - Misinformation Harms | 20.91 | Attack Success Rate (%) | 56.5 |
Interactive version: theaggregate.ai/model?slug=llama-sea-lion-v3-8b-it · How It Works · Data refreshed daily, snapshot 2026-11-22.