Avalon-ToM-Bench - Tacit Coordination: leaderboard
Avalon-ToM-Bench (arXiv 2608.09638v1, Table 1): theory of mind in The Resistance: Avalon, where models judge perspective-constrained true/false statements about players' knowledge and motives from game state and public history. Accuracy on Tacit Coordination (motivational action) items; balanced true/false, chance 50%. Higher is better.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 92.14 | #159 |
| 2 | GPT-5 | 91.43 | #105 |
| 3 | GPT-5 Mini | 89.29 | #193 |
| 4 | Gemini 2.5 Flash | 88.41 | #255 |
| 5 | Qwen 3 8B (Thinking) | 87.4 | #678 (Qwen 3 8B) |
| 6 | Qwen 3 4B 2507 (Thinking) | 84.89 | #527 |
| 7 | Qwen 3 14B (Reasoning) | 83.45 | #542 (Qwen 3 14B) |
| 8 | Phi-4 | 80 | #707 |
| 9 | Qwen 3.5 9B (Thinking) | 79.7 | #391 (Qwen 3.5 9B) |
| 10 | Qwen 3 4B (Reasoning) | 78.03 | #834 (Qwen 3 4B) |
| 11 | GPT-OSS-120B | 77.86 | #346 |
| 12 | GPT-4o Mini | 77.86 | #598 |
| 13 | Gemma 3 27B (IT) | 77.14 | #528 |
| 14 | Mistral 7B Instruct (v0.3) | 73.57 | #1255 |
| 15 | Phi-3.5-mini-instruct | 72.86 | #1148 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=avalon-tom-bench-tacit-coordination · How It Works · Data refreshed daily, snapshot 2026-11-05.