SEA-HELM (Vietnamese) - NLG: leaderboard
SEA-HELM's natural language generation competency for Vietnamese: combines its summarization and translation tasks. Normalized 0-100 score from AI Singapore's SEA-HELM leaderboard: (score - random baseline) / (maximum - baseline), floored at 0 and averaged over eight runs; higher is better.
Metric: Normalized Score. Source: leaderboard.sea-lion.ai. Saturation forecast: Around 2030. 58 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama-SEA-LION-v3-70B-IT | 56.27 | #407 |
| 2 | Gemma 4 26B A4B (IT) (Reasoning) | 55.48 | #293 (Gemma 4 26B A4B (IT)) |
| 3 | Qwen 3.5 122B A10B (Reasoning) | 55.18 | #184 (Qwen 3.5 122B A10B) |
| 4 | Qwen 3.5 27B (Reasoning) | 55.16 | #192 (Qwen 3.5 27B) |
| 5 | Qwen 3.6 35B A3B (Reasoning) | 55.05 | #238 (Qwen 3.6 35B A3B) |
| 6 | Qwen 3.5 35B A3B (Reasoning) | 55.04 | #268 (Qwen 3.5 35B A3B) |
| 7 | Qwen 3.6 27B (Reasoning) | 54.67 | #182 (Qwen 3.6 27B) |
| 8 | Llama 4 Scout Instruct | 54.61 | #547 |
| 9 | Qwen 3 14B (Reasoning) | 54.27 | #542 (Qwen 3 14B) |
| 10 | Qwen 3.5 9B (Reasoning) | 54.24 | #391 (Qwen 3.5 9B) |
| 11 | Llama 3.3 70B Instruct | 54.19 | #541 |
| 12 | Mistral Small 4 | 54.11 | #522 |
| 13 | Qwen 3 32B (Thinking) | 54.1 | #449 (Qwen 3 32B) |
| 14 | Llama-SEA-LION-v3-8B-IT | 53.93 | #800 |
| 15 | Qwen 3 VL 8B Instruct | 53.8 | #419 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sea-helm-vietnamese-nlg · How It Works · Data refreshed daily, snapshot 2026-11-05.