Video-MemComp - EgoSchema: leaderboard
Video-MemComp v1: streaming evaluation of official development sets. Table 1 mixes reproduced starred rows and results from prior papers; frame sampling and average KV budget differ by system and remain in row identity. Backbone checkpoint revision and decoding are not fully specified; source-qualified systems are retained. This is video QA under cache compression, not an unmodified backbone score. EgoSchema. Metric: Accuracy (%).
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Video-MemComp GPT4-o [frames 1fps; KV –] | 72.2 | |
| 2 | Video-MemComp Qwen2-VL-7B + Video-MemComp [frames 768; KV 6K] | 68.6 | |
| 3 | Video-MemComp LongVU 7B [frames 1fps; KV 8K] | 67.6 | |
| 4 | Video-MemComp InternVL3.5-2B + InfiniPot-V [frames 64/100; KV 6K] | 67.6 | |
| 5 | Video-MemComp Qwen2-VL-7B + StreamMem [frames 0.5/0.2fps; KV 6K] | 67.2 | |
| 6 | Video-MemComp Qwen2-VL-7B + InfiniPot-V [frames 768; KV 6K] | 65.6 | |
| 7 | Video-MemComp InternVL3.5-2B + Video-MemComp [frames 64/100; KV 6K] | 65.6 | |
| 8 | Video-MemComp Qwen2-VL-7B [frames 768; KV 100K] | 65.2 | |
| 9 | Video-MemComp Qwen2.5-VL-3B [frames 768; KV 100K] | 64.4 | |
| 10 | Video-MemComp InternVL3.5-2B [frames 8/16/32/64; KV 16K] | 64.4 |
Interactive version: theaggregate.ai/benchmark?slug=video-memcomp-egoschema · How It Works · Data refreshed daily, snapshot 2026-10-27.