Video-MemComp Allocation 0.90 - VideoMME Short: 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. Section 4.4: Qwen2.5-VL-3B with identical intra-layer compression; compare static versus dynamic allocation at similarity threshold 0.90. Metric: Accuracy (%).
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Video-MemComp Qwen2.5-VL-3B dynamic allocation | 71 | |
| 2 | Video-MemComp Qwen2.5-VL-3B static allocation | 70 |
Interactive version: theaggregate.ai/benchmark?slug=video-memcomp-allocation-0-90-videomme-short · How It Works · Data refreshed daily, snapshot 2026-10-27.