Absurd World - Zero-Shot - REAL: leaderboard
Absurd World (Albright et al., 2026) zero-shot Determine Outcome task: from a text commentary of a five-shot-per-team penalty shootout under the REAL ruleset, which mirrors ordinary shootout scoring (a hit scores, the higher total wins), the model names the winner. Accuracy over 100 generated prompts per model, in percent; higher is better.
Metric: Determine-Outcome Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 100 | #452 |
| 2 | GPT-4.1 Mini | 100 | #775 |
| 3 | Qwen 2.5 72B Instruct | 100 | #1093 |
| 4 | Llama 3.1 405B Instruct | 100 | #1114 |
| 5 | DeepSeek R1 0528 | 100 | #425 |
| 6 | Llama 4 Maverick Instruct | 100 | #1125 |
| 7 | Hermes-3-Llama-3.1-405B | 100 | #1022 |
| 8 | Llama 3.1 Nemotron Ultra 253B v1 | 100 | #1027 |
| 9 | GPT-4o Mini | 99 | #1627 |
| 10 | Gemini 2.0 Flash (001) | 95 | #724 |
| 11 | Claude 3.5 Haiku | 94 | #1563 |
| 12 | Gemma 2 27B (IT) | 81 | #1815 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=absurd-world-zero-shot-real · How It Works · Data refreshed daily, snapshot 2026-11-12.