Perception or Prejudice: leaderboard
Multimodal Large Language Models (MLLMs) are increasingly deployed in human-facing roles where personality perception is critical, yet existing benchmarks evaluate this capability solely on numerical Big Five score prediction, leaving open whether models truly perceive personality through behavioral understanding or merely prejudge through superficial pattern matching. We address this gap with thr
Metric: HR (self-reported). Source: benchmarklist.com. Saturation forecast: Around April 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 33.5 | #99 |
| 2 | GPT-5.5 | 28 | #31 |
| 3 | Gemini 3.1 Pro (Preview) | 27.4 | #58 |
| 4 | Gemini 2.5 Pro | 20 | #152 |
| 5 | GPT-5.4 | 17.9 | #79 |
| 6 | Gemini 2.5 Flash | 16.9 | #245 |
| 7 | Claude Opus 4.6 | 16.8 | #63 |
| 8 | Qwen 3.5 397B A17B | 15.9 | #146 |
| 9 | Qwen 3 VL 235B A22B | 12.8 | #238 |
| 10 | Claude Sonnet 4.6 | 12.5 | #90 |
| 11 | Claude Haiku 4.5 | 12.4 | #280 |
| 12 | Gemma 4 31B (IT) | 11.3 | #176 |
| 13 | GPT-5.4 Mini | 11.3 | #212 |
| 14 | O4 Mini | 7.8 | #180 |
| 15 | Llama 4 Maverick | 5.8 | #464 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=perception-or-prejudice · How It Works · Data refreshed daily, snapshot 2026-10-20.