Omni-Persona - Image-to-Image: leaderboard

Image-to-Image persona matching on Omni-Persona (2026): a query face image is matched to profile images to answer personalized questions, scored by calibrated accuracy, which combines judged answers on answerable items with correct abstention on items whose persona is absent from the context. 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), on the image-to-image persona matching group (query face image, profile image context), zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around 2031. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.6 FlashMedium38.5#81
2Qwen3 Omni 30B A3B Instruct37.2#930
3Qwen2.5-Omni-3B35.9#3519
4Gemini 3.5 Flash Lite35.1#299
5gemma-4-E4B-it34.6#1539
6Qwen2.5-Omni-7B33.8#1884
7MiniCPM-o-4.5Thinking31.6#1252
8Gemma 4 E2B (IT)26.4#2451
9Phi-4 Multimodal Instruct26#4154
10Interactive-Omni-8B20.8

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=omni-persona-image-to-image · How It Works · Data refreshed daily, snapshot 2026-11-21.