Why Is Video Still So Expensive? A Survey of Inference-Efficiency Mechanisms in Video and Audiovisual LLMs

Fri, 11 Sep 2026 00:00:00 +0900

Paper

Why Are Video LLMs Still So Expensive? — A Four-Stage Guide to Inference Efficiency Mechanisms

TL;DR — This paper classifies 125 VideoLLM inference efficiency studies published between late 2022 and August 2026 into 4 stages of the encoder–connector–LLM pipeline, and performs controlled comparisons only under “the same host model · the same input protocol,” concluding that accuracy is nearly preserved even when only about 25% of visual tokens remain. (source: Abstract, §I)

JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation

Wed, 02 Jul 2025 00:00:00 +0900

Paper

One-line summary (TL;DR)

JanusFlow achieves state-of-the-art performance in both image understanding and generation (FID 9.51 / GenEval 0.63 / MMBench 74.9) with a single 1.3 B-parameter model built on Rectified Flow + decoupled vision encoders + representation alignment.

Vision-Language Model on Jaehun's Blog

Why Is Video Still So Expensive? A Survey of Inference-Efficiency Mechanisms in Video and Audiovisual LLMs

Why Are Video LLMs Still So Expensive? — A Four-Stage Guide to Inference Efficiency Mechanisms

JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation

One-line summary (TL;DR)