SEA LLM Leaderboard - SeaBench: leaderboard
SeaBench private split aggregate from the LLM Leaderboard for Southeast Asian Languages, measuring open-ended chat quality in Indonesian, Thai, and Vietnamese.
Metric: Private Average Score (0-10). Source: huggingface.co. Saturation forecast: Estimated already saturated. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 | 8.9 | #248 |
| 2 | Claude 3.5 Sonnet (20241022) | 8.78 | #291 |
| 3 | DeepSeek V3 (0324) | 8.78 | #340 |
| 4 | Claude 3.7 Sonnet (20250219) | 8.77 | #201 |
| 5 | GPT-4o (2024-08-06) | 8.64 | #336 |
| 6 | Gemini 1.5 Pro | 8.62 | #445 |
| 7 | GPT-4.1 Mini | 8.6 | #355 |
| 8 | Gemini 2.0 Flash | 8.6 | #342 |
| 9 | Gemma 3 27B (IT) | 8.43 | #523 |
| 10 | Gemini 2.0 Flash Lite | 8.26 | #450 |
| 11 | GPT-4o Mini (2024-07-18) | 8.25 | #558 |
| 12 | Gemini 1.5 Flash | 8.18 | #618 |
| 13 | Gemma 3 12B (IT) | 8.1 | #664 |
| 14 | Qwen 2.5 72B Instruct | 7.96 | #454 |
| 15 | Claude 3.5 Haiku (20241022) | 7.87 | #588 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sea-llm-leaderboard-seabench · How It Works · Data refreshed daily, snapshot 2026-10-20.