OpenEvals - Humanity's Last Exam — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 29 models tracked.

Top models

#ModelScore
1Kimi K2.550.2
2DeepSeek V3.240.8
3GLM-530.5
4Qwen 3.5 397B A17B28.7
5Qwen 3.5 122B A10B25.3
6LongCat-Flash-Thinking-260125.2
7GLM-4.724.8
8Qwen 3.5 27B24.3
9Kimi K2 (Thinking)23.9
10Step 3.5 Flash23.1
11Qwen 3.5 35B A3B22.4
12MiniMax-M2.122.2
13MiMo-V2-Flash22.1
14MiniMax-M2.519.4
15NVIDIA-Nemotron-3-Super-120B-A12B-BF1618.26

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