Balrog: leaderboard
Balanced reasoning and logic games evaluation benchmark. (Source: benchmarklist.com, self-reported.)
Metric: Score (imported). Source: benchmarklist.com. Saturation forecast: Around January 2027. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4 | 43.6 | #181 |
| 2 | Gemini 2.5 Pro (Preview 06-05) | 43.3 | #106 |
| 3 | DeepSeek R1 | 34.9 | #251 |
| 4 | GPT-5.2 | 32.8 | #110 |
| 5 | Claude 3.7 Sonnet | 32.6 | #249 |
| 6 | GPT-4o | 32.3 | #343 |
| 7 | Grok 3 Mini | 29.5 | #281 |
| 8 | Llama 3.1 405B | 27.9 | #466 |
| 9 | Llama 3.3 70B | 23 | #579 |
| 10 | Gemini 1.5 Flash | 21 | #618 |
| 11 | DeepSeek V3 | 19.5 | #321 |
| 12 | Claude 3.5 Haiku | 19.3 | #568 |
| 13 | Mistral Large | 17.6 | #710 |
| 14 | GPT-4o Mini (2024-07-18) | 17.4 | #558 |
| 15 | Qwen 2.5 Max | 16.2 | #416 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=balrog · How It Works · Data refreshed daily, snapshot 2026-10-20.