Multidisciplinary reasoning
Humanity's Last Examwith tools
Humanity's Last Exam — extremely hard expert questions across many subjects. “With tools” means the AI is allowed to search the web or run code while answering. Higher is better.
Rankings
Higher is better164.5%
Claude Fable 5
Anthropic · Jun 9 2026
262.1%
Muse Spark 1.1
Meta · Jul 9 2026
358.7%
GPT-5.4-Pro
OpenAI · Mar 5 2026
457.9%
Claude Opus 4.8
Anthropic · May 28 2026
557.4%
Claude Sonnet 5
Anthropic · Jun 30 2026
657.2%
GPT-5.5-Pro
OpenAI · Apr 23 2026
756%
Kimi K3
Moonshot AI · Jul 16 2026
854.7%
Claude Opus 4.7
Anthropic · Apr 16 2026
854.7%
GLM-5.2
Z.ai · Jun 16 2026
1053%
Claude Opus 4.6
Anthropic · Feb 5 2026
1152.3%
GLM-5.1
Z.ai · Apr 7 2026
1252.2%
GPT-5.5
OpenAI · Apr 23 2026
1351.4%
Gemini 3.1 Pro
Google · Feb 19 2026
1450.4%
GLM-5
Z.ai · Feb 12 2026
1450.4%
Muse Spark
Meta · Apr 8 2026
1642.8%
GLM-4.7
Z.ai · Dec 22 2025