ProactiveBench - VSOD: leaderboard
Metric: Trajectory accuracy (%) on the temporally occluded video frames of VSOD: multi-turn multiple choice in which the model may name a category, abstain, or pick a proactive suggestion that yields a new frame, and a trajectory counts only when it ends in the correct category; samples a first-turn guess solves for at least 25% of the evaluated models are filtered out; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-38B | 31 | #395 |
| 2 | InternVL3-78B | 16.7 | #345 |
| 3 | O4 Mini | 16.7 | #172 |
| 4 | Phi-4 Multimodal Instruct | 16.7 | #896 |
| 5 | InternVL3-8B | 11.9 | #606 |
| 6 | Qwen 2.5 VL 32B Instruct | 4.8 | #443 |
| 7 | Qwen 2.5 VL 72B Instruct | 2.4 | #364 |
| 8 | GPT-5.2 | 0 | #105 |
| 9 | GPT-4.1 | 0 | #240 |
| 10 | Qwen 2.5 VL 7B Instruct | 0 | #643 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=proactivebench-vsod · How It Works · Data refreshed daily, snapshot 2026-10-11.