HELM SeaHELM - XNLI (vi): leaderboard
HELM SeaHELM run of XNLI, Vietnamese natural-language inference (exact match).
Metric: EM. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-11-20) | 68.9 | #381 |
| 2 | Gemini 2.0 Flash | 65.9 | #342 |
| 3 | Claude 3.7 Sonnet (20250219) | 65.8 | #201 |
| 4 | GPT-4o Mini (2024-07-18) | 65.4 | #558 |
| 5 | Qwen 2.5 72B Instruct | 65.1 | #454 |
| 6 | Gemini 2.0 Flash Lite | 65 | #450 |
| 7 | Qwen 2 72B Instruct | 64.1 | #576 |
| 8 | Llama 3.1 405B Instruct | 60.8 | #457 |
| 9 | Llama 3.3 70B Instruct | 59.7 | #534 |
| 10 | Llama 3.1 8B CPT SEA LIONv3 Instruct | 57.9 | #810 |
| 11 | Llama 4 Maverick Instruct FP8 | 30.7 | #414 |
| 12 | Gemma 2 9B (IT) | 25.1 | #788 |
| 13 | Gemma 2 27B (IT) | 23.1 | #645 |
| 14 | Llama 4 Scout Instruct | 21.2 | #539 |
| 15 | DeepSeek R1 Distill Llama 8B | 11.2 | #1291 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-seahelm-xnli-vi · How It Works · Data refreshed daily, snapshot 2026-10-20.