Llama-SEA-LION-v3-70B-IT: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1618 ± 18, rank #377 of 1632 rated models, from 152 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| FilBench - Cultural Knowledge | 76.78 | Category Score (%) | 100 |
| SEA-HELM (Thai) - SEA-Safeguard | 77.46 | Normalized Score | 100 |
| SEA-HELM (Vietnamese) - Summarization | 19.81 | Normalized Score | 100 |
| SEA-HELM (Indonesian) - Summarization | 19.14 | Normalized Score | 98.2 |
| SEA-HELM (Tamil) - Summarization | 14.12 | Normalized Score | 98.2 |
| SEA-HELM (Thai) - Summarization | 25.38 | Normalized Score | 98.2 |
| SEA-HELM (Vietnamese) - NLG | 56.27 | Normalized Score | 98.2 |
| FilBench - Classical NLP | 89.99 | Category Score (%) | 97.2 |
| SEA-HELM (Indonesian) - NLG | 56.02 | Normalized Score | 96.5 |
| ThaiSafetyBench - Human-Chatbot Interaction Harms | 12.39 | Attack Success Rate (%) | 95.7 |
| SEA-HELM (Filipino) - NLG | 57.42 | Normalized Score | 93 |
| SEA-HELM (Filipino) - Summarization | 25.04 | Normalized Score | 93 |
Interactive version: theaggregate.ai/model?slug=llama-sea-lion-v3-70b-it · How It Works · Data refreshed daily, snapshot 2026-10-08.