BALROG Crafter (VLM) — leaderboard

Metric: Progress (%). Source: balrogai.com. 10 models tracked.

Top models

#ModelScore
1Human Expert90
2Claude 3.5 Sonnet (20241022)37.3
3Gemini 2.5 Pro (03-25)37.3
4Gemini 1.5 Pro (002)33.5
5GPT-4o (2024-05-13)26.8
6Gemini 1.5 Flash (002)20.7
7GPT-4o Mini (2024-07-18)19.9
8Llama 3.2 11B Instruct15.9
9Llama 3.2 90B Instruct14.5

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