HELM TORR - Table Scenarios: leaderboard

Metric: Mean score. Source: crfm.stanford.edu. 14 models tracked.

Top models

#ModelScore
1Claude 3.5 Sonnet (20241022)49.93
2GPT-4o (2024-11-20)49.75
3DeepSeek V349.73
4Gemini 1.5 Pro (002)47.96
5Llama 3.1 405B Instruct46.12
6Qwen 2 72B Instruct45.41
7Gemini 1.5 Flash (002)45.37
8Llama 3.1 70B Instruct44.36
9GPT-4o Mini (2024-07-18)43.08
10Claude 3.5 Haiku (20241022)42.69
11Mistral 7B Instruct (v0.3)32.42
12Llama 3.1 8B Instruct28.85

Interactive version: theaggregate.ai/benchmark?slug=helm-torr-table-scenarios · How It Works · Data refreshed daily, snapshot 2026-09-08.