Thai LLM NLG - iapp_squad_seacrowd_qa (ROUGE2): leaderboard
The Thai LLM Leaderboard, run as an open results dataset by the Thai LLM community. Generation quality on iapp-SQuAD Thai extractive question answering, scored by ROUGE-2 (SEACrowd harness).
Metric: ROUGE2. Source: huggingface.co. Saturation forecast: Estimated already saturated. 68 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 59.33 | #534 |
| 2 | Llama 3 8B Cpt Sea Lionv2.1 Instruct | 59.15 | #1024 |
| 3 | Llama 3.1 70B Cpt Sea Lionv3 Instruct | 58.31 | #548 |
| 4 | Llama 3.1 8B Instruct | 55.69 | #1035 |
| 5 | Llama 3.1 70B Instruct | 52.8 | #561 |
| 6 | Llama 3.1 Nemotron 70B Instruct | 50.74 | #554 |
| 7 | Llama 3.1 8B CPT SEA LIONv3 Instruct | 50.73 | #810 |
| 8 | Llama 3.2 3B Instruct | 47.74 | #1352 |
| 9 | llama-3-typhoon-v1.5-8B-instruct | 44.22 | #1198 |
| 10 | Llama 3 8B Cpt Sea Lionv2 Instruct | 42.74 | #1188 |
| 11 | SeaLLMs-v3-7B-Chat | 39.36 | #915 |
| 12 | Llama 3.2 1B Instruct | 32.43 | #1539 |
| 13 | GPT-4o Mini (2024-07-18) | 30.63 | #558 |
| 14 | gemma2-9B-cpt-sea-lionv3-instruct | 29.21 | #705 |
| 15 | Qwen 2 7B Instruct | 29.17 | #991 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=thai-llm-nlg-iapp-squad-seacrowd-qa-rouge2 · How It Works · Data refreshed daily, snapshot 2026-10-20.