Omni-Persona: leaderboard
Omni-Persona (2026) is an evaluation-only benchmark of omnimodal personalization: models ground a query in a user persona given as face images, biography text or voice audio, including items whose persona is absent. The score is calibrated accuracy, which combines judged answers on answerable items with correct abstention on items whose persona is absent from the context, over 18 tasks; percent, higher is better.
Metric: Calibrated accuracy (%): item-weighted accuracy over answerable items (answer judged correct by a GPT-5.4-mini judge) and absent-persona items (correct abstention by keyword matching), over all four matching scenario groups and 18 tasks, zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around 2034. 10 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | MiniCPM-o-4.5 | Thinking | 39.3 | #1251 |
| 2 | Qwen2.5-Omni-3B | 36.7 | #3416 | |
| 3 | Gemini 3.6 Flash | Medium | 35.7 | #84 |
| 4 | Qwen3 Omni 30B A3B Instruct | 33.7 | #928 | |
| 5 | Phi-4 Multimodal Instruct | 33.2 | #4155 | |
| 6 | Gemma 4 E2B (IT) | 32.1 | #2532 | |
| 7 | gemma-4-E4B-it | 31.2 | #1577 | |
| 8 | Qwen2.5-Omni-7B | 30.9 | #1888 | |
| 9 | Gemini 3.5 Flash Lite | 29.7 | #308 | |
| 10 | Interactive-Omni-8B | 29.6 |
No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=omni-persona · How It Works · Data refreshed daily, snapshot 2026-11-22.