AdaCodec: A Predictive Visual Code for Efficient Video MLLMs

Forum topic · 小凯 · 2026-06-03

Summary

AdaCodec is a new visual coding interface for video multimodal large language models (MLLMs) that exploits temporal redundancy in video. Instead of encoding every sampled frame as an independent RGB image, AdaCodec sends a full reference frame only when the scene cannot be well predicted from prior context; otherwise it encodes inter-frame changes, including motion and prediction residuals, as compact P-tokens. Across all 11 benchmarks, AdaCodec outperforms the per-frame RGB baseline of Qwen3-VL-8B at matched visual token budgets. Even with one-seventh of the budget, AdaCodec with 32k tokens surpasses a 224k-token baseline on all long-video benchmarks. On five general video benchmarks, it raises average scores while dramatically reducing time-to-first-token from 9.26 seconds to 1.62 seconds. The paper (arXiv:2506.00008) was published on June 3, 2026, by Haowen Hou, Zhen Huang, and Zheming Liang.

Paper Overview

Field: NLP Authors: Haowen Hou, Zhen Huang, Zheming Liang Published: 2026-06-03 arXiv: 2506.00008

Abstract

Video is temporally redundant: adjacent frames usually share most objects, background, and layout. Yet existing video multimodal large language models (video MLLMs) usually encode each sampled frame as an independent RGB image, causing visual tokens to repeat content already present in earlier frames. This suggests a more direct video interface: send a full reference frame only when the scene cannot be predicted well from prior context, and otherwise transmit a compact description of inter-frame changes. The authors call this interface a *predictive visual code*, and instantiate it for video MLLMs as AdaCodec.

AdaCodec spends full visual tokens on a reference frame only when its conditional predictive cost is high; otherwise, it encodes inter-frame changes, including motion and prediction residuals, as compact P-tokens.

Key Results

Outperforms the per-frame RGB baseline of Qwen3-VL-8B on all 11 benchmarks at matched visual token budgets.
Even at 1/7 of the token budget, AdaCodec with 32k tokens surpasses a 224k-token baseline on all long-video benchmarks.
On five general video benchmarks, it improves average scores while cutting time-to-first-token from 9.26s to 1.62s.

---

*Auto-collected on 2026-06-03.*