LiveBench Plot Unscrambling — leaderboard

Metric: Score. Source: livebench.ai. 73 models tracked.

Top models

#ModelScore
1Claude 3.5 Sonnet (20240620)55.14
2Claude 3 Opus (20240229)50.16
3Llama 3.1 405B Instruct47.38
4GPT-4 (0613)45.54
5GPT-4o (2024-08-06)45.11
6GPT-4 Preview (0125)43.49
7Gemini 1.5 Pro (Preview 0827)43.44
8GPT-4o (2024-05-13)42.81
9GPT-4o ChatGPT42.53
10Gemini 1.5 Pro (Preview 0801)41.87
11Mistral Large 2 (Jul)40.05
12Gemini 1.5 Pro (0514)39.92
13GPT-4 Turbo39.62
14Llama 3.1 70B Instruct39.42
15Hermes 3 - Llama-3.1 70B37.32

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