M3-VQA (Gold Entity Names): leaderboard
Metric: IoU accuracy (%) between the predicted and gold answer sets (exact match against Wikidata answer aliases, one-year margin for dates), averaged over the M3-VQA multi-entity, multi-hop knowledge-based visual questions, with the names of the entities behind the gold evidence of every hop added to the input; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 51.21 |
| 2 | Qwen 2.5 VL 72B Instruct | 45.05 |
| 3 | InternVL2.5-78B | 45.02 |
| 4 | Qwen 2.5 VL 32B Instruct | 40.39 |
| 5 | MiniCPM-V-2.6 | 33.6 |
| 6 | Qwen 2.5 VL 7B Instruct | 31.83 |
| 7 | Qwen 2 VL 7B Instruct | 30 |
| 8 | InternVL2.5-2B | 23.2 |
Interactive version: theaggregate.ai/benchmark?slug=m3-vqa-gold-entity-names · How It Works · Data refreshed daily, snapshot 2026-10-07.