BALROG Official VLM Leaderboard: leaderboard
Official BALROG VLM leaderboard for long-horizon game reasoning across BabyAI, Crafter, BabaIsAI, MiniHack, NetHack, and related game environments.
Metric: % Progress (percentage). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 35.7 | #123 |
| 2 | Claude 3.5 Sonnet (20241022) | 35.5 | #291 |
| 3 | Gemini 1.5 Pro (002) | 25.8 | #349 |
| 4 | GPT-4o (2024-05-13) | 22.6 | #325 |
| 5 | Llama 3.2 90B Instruct | 21 | #601 |
| 6 | GPT-4o Mini (2024-07-18) | 15.4 | #558 |
| 7 | Gemini 1.5 Flash (002) | 14.9 | #583 |
| 8 | Llama 3.2 11B Instruct | 8.4 | #1126 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=balrog-official-vlm-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.