MirrorBench - Level 0 (Guided Mirror Perception): 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. Level 0, where the mirror is announced and a chain-of-thought template is given. 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) at Level 0, where the mirror is announced and a chain-of-thought template is given, over 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 January 2027. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Pro0.705#280
2Claude Sonnet 4.50.671#267
3Gemini 2.5 Flash0.618#466
4GLM-4.5V0.535#756
5GPT-4.10.51#506
6GPT-4o0.476#754
7Qwen 2.5 VL 72B0.334#782
8Qwen-VL-Max (MirrorBench checkpoint unspecified)0.331
9GPT-4o Mini0.086#1680
10Qwen 2.5 VL 32B0.033#944
11InternVL3.5-1B-0.068#6140
12InternVL3.5-4B-0.243#1946
13LLaVA-v1.5-7B-0.309#7642
14InternVL3.5-2B-0.32#3313
15LLaVA-1.6-7B-0.428#4990

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mirrorbench-level-0-guided-mirror-perception · How It Works · Data refreshed daily, snapshot 2026-11-22.