SVFSearch: leaderboard
Multimodal large language models are increasingly used as agent backbones that understand multimodal inputs, plan retrieval actions, invoke external tools, and reason over retrieved information. Yet existing benchmarks rarely evaluate this ability in short-video applications, where a paused frame is often visually ambiguous and answering requires vertical, long-tail, and fast-evolving domain knowl
Metric: Overall Acc. (self-reported). Source: benchmarklist.com. 74 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3.5-27B | 95.4 | |
| 2 | Qwen3.5-27B [Oracle] | 95.4 | |
| 3 | Qwen3-VL-32B | 90.8 | |
| 4 | Qwen3-VL-32B [Oracle] | 90.8 | |
| 5 | Qwen3.5-9B | 90.3 | |
| 6 | Qwen3.5-9B [Oracle] | 90.3 | |
| 7 | Qwen3.5-4B | 87.9 | |
| 8 | Qwen3.5-4B [Oracle] | 87.9 | |
| 9 | Qwen3-VL-8B [Oracle] | 86.5 | |
| 10 | Qwen3-VL-4B | 84.6 |
Interactive version: theaggregate.ai/benchmark?slug=svfsearch · How It Works · Data refreshed daily, snapshot 2026-10-20.