Atlas

Benchmarks

← All benchmarks

Humanity's Last Exam (Preview)

General QA · 2025-01-24

Preview score on Humanity's Last Exam expert-level questions across science, math, and humanities.

Top models (higher is better)

ModelScore
Gemini 2.5 Pro Preview 03-2518.8
O1 Pro9.2
Claude 3.7 Sonnet8.9
O18.8
Gemini 2.0 Flash Thinking Experimental 01-217.2
Gemini 2.0 Pro Experimental 02-057.1
Llama 4 Maverick Instruct6.8
GPT-4.56.4
Llama 3.2 90B Vision Instruct5.5
Gemini 1.5 Pro 0025.2
Gemini 2.0 Flash Experimental5.2
Gemini 2.0 Flash 0015.1
Claude 3.5 Sonnet (Oct 2024)4.8
Qwen2 VL 72B Instruct4.7
Nova Pro4.6
Gemini 2.0 Flash-Lite4.6
Opus 34.2
Gemini 1.5 Flash 0024.2
Nova Lite4.0
GPT-4o (2024-11-20)3.1
Loading Atlas data…