SeaEval - Cultural Reasoning - SG-Eval v2 Open (Zero-Shot): leaderboard
SeaEval zero-shot SG-Eval v2 Open evaluation measuring culturally grounded reasoning and regional knowledge for Southeast Asian and comparison contexts.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2 7B Instruct | 56.56 | #991 |
| 2 | SeaLLMs-v3-7B-Chat | 55 | #915 |
| 3 | gemma2-9B-cpt-sea-lionv3-instruct | 55 | #705 |
| 4 | Sailor2-8B-Chat | 54.36 | #782 |
| 5 | Qwen 2 72B Instruct | 54.08 | #576 |
| 6 | Gemma 2 9B (IT) | 53.96 | #788 |
| 7 | Qwen 2.5 72B Instruct | 53.32 | #454 |
| 8 | Qwen 2.5 32B Instruct | 53.2 | #498 |
| 9 | Qwen 2.5 14B Instruct | 53.2 | #647 |
| 10 | Gemma 2 2B (IT) | 52.08 | #1341 |
| 11 | Llama 3.1 70B Cpt Sea Lionv3 Instruct | 51.92 | #548 |
| 12 | Llama 3.1 70B Instruct | 51.32 | #561 |
| 13 | Llama 3 8B Instruct | 51.12 | #1129 |
| 14 | Llama 3 70B Instruct | 50.6 | #634 |
| 15 | Qwen 2.5 7B Instruct | 50.28 | #862 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=seaeval-cultural-reasoning-sg-eval-v2-open-zero-shot · How It Works · Data refreshed daily, snapshot 2026-10-20.