MirrorBench - Robot Body: 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. Score with robotic bodies and hands, over four prompt levels. The score averages task success, step-wise improvement and final and peak completion ratios (at most 1; the final completion ratio is negative when the hand ends farther away than it started); higher is better.

Metric: Average score (mean of task success rate, step-wise improvement ratio, final completion ratio and peak completion ratio, each at most 1; the final completion ratio is negative when the hand ends farther from the mark than it started) with the robotic bodies and hands, over the four prompt levels, 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

#ModelScoreOverall rank
1Gemini 2.5 Pro0.362#276
2Gemini 2.5 Flash0.218#463
3GPT-4.10.035#505
4Claude Sonnet 4.5-0.013#261
5GPT-4o-0.014#751
6Qwen-VL-Max (MirrorBench checkpoint unspecified)-0.032
7Qwen 2.5 VL 72B-0.038#781
8Qwen 2.5 VL 32B-0.056#941
9InternVL3.5-4B-0.066#1930
10GLM-4.5V-0.174#754
11LLaVA-Next-LLaMA3-8B-0.288#5654
12GPT-4o Mini-0.352#1667
13LLaVA-1.6-7B-0.372#4955
14InternVL3.5-1B-0.433#6112
15Qwen 2.5 VL 7B-0.466#1862

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mirrorbench-robot-body · How It Works · Data refreshed daily, snapshot 2026-11-21.