CausalGame - Prompting Mode: leaderboard
The CausalGame scenarios with the model writing Python code blocks that call the game's API directly, instead of making structured tool calls in a ReAct loop. Mean survival rate over 14 scenarios and three seeds. Higher is better. Source: arxiv.org/abs/2607.04293.
Metric: Mean Survival Rate (%; 14 scenarios, 3 seeds). Source: arxiv.org. Saturation forecast: Around 2030. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 68.6 | #53 |
| 2 | GLM-5.2 | 68.3 | #82 |
| 3 | Claude Opus 4.5 | 67.6 | #90 |
| 4 | Qwen 3.7 Max | 67.4 | #88 |
| 5 | Grok 4.1 Fast | 66.3 | #227 |
| 6 | Kimi K2.6 | 66.2 | #115 |
| 7 | DeepSeek V4 Pro | 65.2 | #113 |
| 8 | GLM-5.1 | 65.2 | #111 |
| 9 | Kimi K2.5 | 64.5 | #154 |
| 10 | Claude Sonnet 4.5 | 63.7 | #150 |
| 11 | Gemini 3.5 Flash | 62.9 | #68 |
| 12 | Hy3-preview | 60.8 | #168 |
| 13 | MiMo-V2.5-Pro | 60.3 | #169 |
| 14 | GPT-5.5 | 60.1 | #34 |
| 15 | MiniMax-M2 | 60 | #329 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #6 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=causalgame-prompting-mode · How It Works · Data refreshed daily, snapshot 2026-11-07.