MirrorBench - Task Success Rate: leaderboard
MirrorBench (2026) adapts the mirror self-recognition test to multimodal models: in a simulated scene the model moves a hand to touch a mark on its own body that it can only see in a mirror. Task success rate, the share of episodes in which the hand reaches the mark, over all levels and bodies, from 0 to 1; higher is better.
Metric: Task success rate (0-1), the share of episodes in which the hand reaches the mark, over the four prompt levels and both body settings, on MirrorBench's Isaac Sim mirror self-recognition scenes (7 human and robot bodies, 6 hands, 6 marks; the agent moves its hand with six directional actions to touch a mark on its own body seen in a mirror; step budget the shortest path plus 10), temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 18 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 0.17 | #261 |
| 2 | Gemini 2.5 Pro | 0.155 | #276 |
| 3 | GPT-4.1 | 0.115 | #505 |
| 4 | GLM-4.5V | 0.096 | #754 |
| 5 | Gemini 2.5 Flash | 0.064 | #463 |
| 6 | GPT-4o | 0.061 | #751 |
| 7 | Qwen-VL-Max (MirrorBench checkpoint unspecified) | 0.053 | |
| 8 | Qwen 2.5 VL 72B | 0.023 | #781 |
| 9 | GPT-4o Mini | 0.008 | #1667 |
| 10 | LLaVA-Next-LLaMA3-8B | 0.004 | #5654 |
| 11 | InternVL3.5-4B | 0.004 | #1930 |
| 12 | Qwen 2.5 VL 32B | 0.004 | #941 |
| 13 | Qwen 2.5 VL 7B | 0 | #1862 |
| 14 | LLaVA-v1.5-7B | 0 | #7630 |
| 15 | Qwen 2.5 VL 3B | 0 | #2639 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=mirrorbench-task-success-rate · How It Works · Data refreshed daily, snapshot 2026-11-21.