HELM SeaHELM - ViHSD — leaderboard

Metric: Macro F1 score. Source: crfm.stanford.edu. 21 models tracked.

Top models

#ModelScore
1GPT-4o (2024-11-20)61.63
2Claude 3.7 Sonnet (20250219)59.89
3Qwen 2 72B Instruct54.54
4Qwen 2.5 72B Instruct53.98
5Gemini 2.0 Flash51.19
6Llama 3.3 70B Instruct50.54
7Llama 3.1 70B Instruct48.8
8Gemini 2.0 Flash Lite46.06
9Llama 4 Scout Instruct45.15
10Llama 3.1 405B Instruct42.01
11GPT-4o Mini (2024-07-18)41.19
12Llama 4 Maverick Instruct FP839.41
13Llama 3.2 90B Vision Instruct38.29
14Gemma 2 27B (IT)25.62
15DeepSeek R1 Distill Llama 8B15.65

Interactive version: theaggregate.ai/benchmark?slug=helm-seahelm-vihsd · How the rankings work · Data refreshed daily, snapshot 2026-07-22.