MMBench-Live - Relation Reasoning: leaderboard
Metric: Accuracy (%; relation reasoning questions; 5.9K newly generated evaluation instances aligned with the MMBench task hierarchy; temperature 0). Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 VL 7B Instruct | 85.71 |
| 2 | Qwen 3 VL 8B Instruct | 84.42 |
Interactive version: theaggregate.ai/benchmark?slug=mmbench-live-relation-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-29.