BALROG Crafter (VLM): leaderboard
Vision-language models on Crafter survival game using rendered screenshots, measuring visual interpretation for resource management and multi-step goals.
Metric: Progress (%). Source: balrogai.com. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 37.27 | #123 |
| 2 | Claude 3.5 Sonnet (20241022) | 37.27 | #291 |
| 3 | Gemini 1.5 Pro (002) | 33.5 | #349 |
| 4 | GPT-4o (2024-05-13) | 26.81 | #325 |
| 5 | Gemini 1.5 Flash (002) | 20.7 | #583 |
| 6 | GPT-4o Mini (2024-07-18) | 19.91 | #558 |
| 7 | Llama 3.2 11B Instruct | 15.91 | #1126 |
| 8 | Llama 3.2 90B Instruct | 14.54 | #601 |
| 9 | Qwen 2 VL 7B Instruct | 5.45 | #826 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=balrog-crafter-vlm · How It Works · Data refreshed daily, snapshot 2026-10-20.