SocialPersona - Direct Profiling: leaderboard
SocialPersona (2026) tests whether multimodal models can infer what a person cares about from a real social-media timeline (post text, image captions and timestamps) and use it. This board is interest-tag F1 in percent against human-verified stable and recent interests when the model profiles the whole timeline in one pass; higher is better.
Metric: Interest-tag F1 (%) of the constructed profile against human-verified stable and recent interest tags in seven domains (exact match, then o3 semantic matching), from timelines of 100 social-media users given as post text, image captions and timestamps (up to 200 posts), the whole timeline in one pass; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 33.38 | #123 |
| 2 | Gemini 2.5 Flash | 32.88 | #453 |
| 3 | GPT-4o Mini | 27.85 | #1623 |
| 4 | Qwen 3.5 35B A3B | 26.69 | #482 |
| 5 | Qwen 3 VL 8B Instruct | 22.46 | #981 |
| 6 | Qwen 2.5 VL 7B Instruct | 1.67 | #1833 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=socialpersona-direct-profiling · How It Works · Data refreshed daily, snapshot 2026-11-13.