HELM CLEVA - Cleva Scenarios: leaderboard
Metric: Mean win rate. Source: crfm.stanford.edu. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4 (0613) | 95.56 |
| 2 | GPT-3.5 Turbo (0613) | 60 |
| 3 | LLaMA-65B | 35.56 |
Interactive version: theaggregate.ai/benchmark?slug=helm-cleva-cleva-scenarios · How It Works · Data refreshed daily, snapshot 2026-09-08.