HellaSwag — leaderboard

Commonsense natural language inference benchmark with 70K questions. Given a scenario, models must pick the most plausible continuation. Heavily saturated by modern models.

Metric: Accuracy (%). Source: epoch.ai. Status: saturation imminent. 76 models tracked.

Top models

#ModelScore
1Median Human95.6
2GPT-4 (0314)95.3
3Llama 3.1 405B89.2
4falcon-180B89
5DeepSeek V388.9
6DeepSeek V287.1
7Mixtral 8x7B (v0.1)86.7
8text-davinci-00385.5
9Llama 2 70B Base85.3
10falcon-40B85.28
11Qwen 2.5 72B84.8
12LLaMA-65B84.2
13StableBeluga284.1
14Qwen 2.5 Coder 32B83
15falcon-11B82.91

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