SeaLLM-7B-v2: benchmark results
Sea AI Lab's SeaLLM-7B-v2, a multilingual model for Southeast Asian languages (English, Chinese, Vietnamese, Indonesian, Thai, Malay, Khmer, Lao, Burmese, Filipino), about half the size of the earlier SeaLLM-13B. Provider: Alibaba. Released 2024-01-29. Access: Open.
Aggregate Rating 1417 ± 17, rank #1160 of 1656 rated models, from 41 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Enkrypt AI - Toxicity Risk | 0.55 | Risk Score | 89.1 |
| Open Korean LLM - Ko-EQ-Bench | 36.7 | EQ-Bench score | 87.2 |
| Open Portuguese LLM - FaQuAD NLI | 78.42 | Macro F1 (%) | 84 |
| Open Korean LLM - KorNAT-Helpful | 50 | Normalized accuracy (%) | 83.5 |
| Enkrypt AI - Safety Risk | 20.32 | Risk Score | 83.2 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.76 | Normalized accuracy (%) | 81.3 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 44.43 | Exact match, flexible extract (%) | 79.6 |
| Open Korean LLM Leaderboard | 40.32 | Average Score (%) | 76.5 |
| Open Ko-LLM Leaderboard | 40.32 | Average (%) | 76.3 |
| Open Korean LLM - Ko-IFEval | 34.92 | Strict accuracy, prompt/instruction mean (%) | 71.7 |
| Enkrypt AI - Insecure Code Risk | 16 | Risk Score | 69.1 |
| Enkrypt AI - Jailbreak Risk | 6.97 | Risk Score | 66.8 |
Interactive version: theaggregate.ai/model?slug=seallm-7b-v2 · How It Works · Data refreshed daily, snapshot 2026-10-20.