Thai LLM NLG - xl_sum_tha_seacrowd_t2t (ROUGE2): leaderboard
The Thai LLM Leaderboard, run as an open results dataset by the Thai LLM community. Generation quality on XL-Sum Thai summarisation, 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 | 14.24 | #534 |
| 2 | Llama 3.1 Nemotron 70B Instruct | 13.9 | #554 |
| 3 | Llama 3.1 70B Instruct | 13.53 | #561 |
| 4 | Llama 3.1 70B Cpt Sea Lionv3 Instruct | 13.41 | #548 |
| 5 | Llama 3.1 8B Instruct | 11.53 | #1035 |
| 6 | Llama 3 8B Cpt Sea Lionv2.1 Instruct | 11.33 | #1024 |
| 7 | Llama 3.2 3B Instruct | 10.86 | #1352 |
| 8 | llama-3-typhoon-v1.5-8B-instruct | 10.4 | #1198 |
| 9 | SeaLLMs-v3-7B-Chat | 10.23 | #915 |
| 10 | Qwen 2.5 32B Instruct | 9.83 | #498 |
| 11 | Llama 3.2 1B Instruct | 9.77 | #1539 |
| 12 | GPT-4o (2024-05-13) | 9.49 | #325 |
| 13 | GPT-4o Mini (2024-07-18) | 9.45 | #558 |
| 14 | Llama 3.1 8B CPT SEA LIONv3 Instruct | 9.31 | #810 |
| 15 | Qwen 2.5 72B Instruct | 9.21 | #454 |
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-xl-sum-tha-seacrowd-t2t-rouge2 · How It Works · Data refreshed daily, snapshot 2026-10-20.