Thai LLM NLU - xnli.tha_seacrowd_pairs: leaderboard
The Thai LLM Leaderboard, run as an open results dataset by the Thai LLM community. Natural-language understanding accuracy on XNLI’s Thai natural-language-inference pairs (SEACrowd harness).
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 70 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-05-13) | 57.54 | #325 |
| 2 | Claude 3.5 Sonnet (20240620) | 55.69 | #299 |
| 3 | GPT-4o Mini (2024-07-18) | 52.06 | #558 |
| 4 | gemma2-9B-cpt-sea-lionv3-instruct | 47.54 | #705 |
| 5 | Qwen 2.5 72B Instruct | 44.65 | #454 |
| 6 | Gemma 2 9B (IT) | 43.93 | #788 |
| 7 | QwQ 32B-Preview | 41.34 | #651 |
| 8 | Qwen 2.5 14B Instruct | 40.88 | #647 |
| 9 | Qwen 2.5 7B Instruct | 40.82 | #862 |
| 10 | Qwen 2 7B Instruct | 39.2 | #991 |
| 11 | Qwen 2 72B Instruct | 37.58 | #576 |
| 12 | Qwen 2.5 32B Instruct | 37.29 | #498 |
| 13 | Sailor2-8B-Chat | 36.99 | #782 |
| 14 | aya-expanse-32B | 36.43 | #692 |
| 15 | Sailor2-20B-Chat | 35.75 | #590 |
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-nlu-xnli-tha-seacrowd-pairs · How It Works · Data refreshed daily, snapshot 2026-10-20.