HELM SeaHELM - ViHSD — leaderboard
Metric: Macro F1 score. Source: crfm.stanford.edu. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o (2024-11-20) | 61.63 |
| 2 | Claude 3.7 Sonnet (20250219) | 59.89 |
| 3 | Qwen 2 72B Instruct | 54.54 |
| 4 | Qwen 2.5 72B Instruct | 53.98 |
| 5 | Gemini 2.0 Flash | 51.19 |
| 6 | Llama 3.3 70B Instruct | 50.54 |
| 7 | Llama 3.1 70B Instruct | 48.8 |
| 8 | Gemini 2.0 Flash Lite | 46.06 |
| 9 | Llama 4 Scout Instruct | 45.15 |
| 10 | Llama 3.1 405B Instruct | 42.01 |
| 11 | GPT-4o Mini (2024-07-18) | 41.19 |
| 12 | Llama 4 Maverick Instruct FP8 | 39.41 |
| 13 | Llama 3.2 90B Vision Instruct | 38.29 |
| 14 | Gemma 2 27B (IT) | 25.62 |
| 15 | DeepSeek R1 Distill Llama 8B | 15.65 |
Interactive version: theaggregate.ai/benchmark?slug=helm-seahelm-vihsd · How the rankings work · Data refreshed daily, snapshot 2026-07-22.