GroupToM-Bench - Belief: leaderboard
Metric: strict exact-match accuracy (%) over multi-select options (random 6.7) on Level 1 belief tracking (second-order and deeper beliefs under information asymmetry and deception), multiple choice; 240 multimodal multi-agent scenarios (scene image plus dialogue); higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 79.8 |
| 2 | Gemini 3 Pro | 78.9 |
| 3 | GPT-5 | 76.7 |
| 4 | Claude Haiku 4.5 | 75.1 |
| 5 | GPT-5 Mini | 70.4 |
| 6 | InternVL3.5-8B | 66.5 |
| 7 | Qwen 2.5 VL 7B Instruct | 65.8 |
| 8 | GPT-5 Nano | 63.3 |
| 9 | Qwen 2 VL 7B | 58.3 |
Interactive version: theaggregate.ai/benchmark?slug=grouptom-bench-belief · How It Works · Data refreshed daily, snapshot 2026-09-29.