Personality Recovery on 6 HEXACO domains (N=290)
0.75Mean r (Sonnet)GPT-4.1
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-4.1Provider=OpenAI, Architecture=Autoregressive, Sample size (N)=290, Scoring protocol=self-scoring excluded2026.04 | 0.75 | 0.738 | 0.728 | 0.739 | |
| Gemini 3 FlashProvider=Google, Architecture=Autoregressive, Sample size (N)=290, Scoring protocol=self-scoring excluded2026.04 | 0.744 | 0.736 | — | 0.74 | |
| Grok 4.1 FastProvider=xAI, Architecture=Autoregressive, Sample size (N)=290, Scoring protocol=self-scoring excluded2026.04 | 0.74 | 0.735 | 0.719 | 0.731 | |
| Mercury 2Provider=Inception Labs, Architecture=Diffusion, Sample size (N)=290, Scoring protocol=self-scoring excluded2026.04 | 0.693 | 0.697 | 0.654 | 0.681 |