PetQA-Bench - Multimodal Factuality: leaderboard
PetQA-Bench with the owner's photograph attached: claim-level factuality of the multimodal answer. Higher is better. Source: arxiv.org/abs/2609.04598.
Metric: Factuality (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 Mini | 53.1 | #364 |
| 2 | GPT-4.1 | 53.1 | #259 |
| 3 | Gemini 2.5 Pro | 52.4 | #159 |
| 4 | Gemma 3 27B | 49.6 | #606 |
| 5 | Gemini 2.5 Flash | 49.5 | #254 |
| 6 | Gemma 3 12B | 49 | #674 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #6 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=petqa-bench-multimodal-factuality · How It Works · Data refreshed daily, snapshot 2026-11-06.