IPIBench - Repeated Proactiveness: leaderboard

IPIBench (2026) streams videos to multimodal LLMs at one frame per second and checks whether they respond proactively at the right moment, manage monitoring tasks the user cancels or changes, and handle reactive questions interleaved with proactive tasks. This board is accuracy on repeated proactiveness (trigger on every recurrence of the event); percent, higher is better.

Metric: Accuracy (%) on repeated proactiveness (trigger on every recurrence of the event), on IPIBench (1,831 videos, 3,738 interaction instances from egocentric and exocentric datasets) under a unified streaming protocol at 1 FPS (offline models see a sliding window of the latest 16 frames); a proactive response counts only if the first trigger falls within one second of the ground-truth moment, multi-turn tasks only if every required trigger and answer is correct, reactive answers by bidirectional substring matching against up to 20 reference answers; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 16 models tracked. Updated 14 Nov 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro18.4#145
2GPT-4o13.68#751
3Gemini 2.5 Pro13.21#276
4Qwen 3.5 9B (Harness: AURA2 IPIBench offline30frames; Thinking: On)13.21
5GLM-4.6V10.85#710
6GPT-5.48.96#135
7MOSS-VL-Realtime8.96
8Qwen 3.5 Plus7.89#242
9Qwen 3 VL 8B4.25#917
10InternVL3-8B3.77#1782
11VideoLLM-online-8B1.82
12Dispider1.82#3815
13Qwen 3.5 9B (Harness: AURA2 IPIBench offline30frames; Thinking: Off)0.94
14JoyAI-VL-Interaction0.47
15LLaVA-OneVision-7B0#2981

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=ipibench-repeated-proactiveness · How It Works · Data refreshed daily, snapshot 2026-11-21.