# OmniFocus: Query-Guided Modality-Balanced Token Compression for Omni-Modal Large Language Models

arXiv: https://arxiv.org/abs/2607.03050 · HTML: https://mlsanity.com/paper/2607.03050/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Audio and omni-modal token compression, one of 12 directions in Efficient LLM inference and serving.
- Builds on 39 of its 43 references that are in the map.
- First posted on arXiv 3 Jul 2026.

## Placed in

- Efficient LLM inference and serving / [Audio and omni-modal token compression](https://mlsanity.com/inference/audio-and-omni-modal-token-compression/)

## Builds on (most cited first, 30 of 39)

- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models](https://mlsanity.com/paper/2403.06764/) (2024)
- [Qwen3-Omni Technical Report](https://mlsanity.com/paper/2509.17765/) (2025)
- [LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models](https://mlsanity.com/paper/2403.15388/) (2024)
- [VisionZip: Longer is Better but Not Necessary in Vision Language Models](https://mlsanity.com/paper/2412.04467/) (2024)
- [Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming](https://mlsanity.com/paper/2408.16725/) (2024)
- [Qwen3.5-Omni Technical Report](https://mlsanity.com/paper/2604.15804/) (2026)
- [Adaptive Keyframe Sampling for Long Video Understanding](https://mlsanity.com/paper/2502.21271/) (2025)
- [WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMs](https://mlsanity.com/paper/2502.04326/) (2025)
- [Fit and Prune: Fast and Training-free Visual Token Pruning for Multi-modal Large Language Models](https://mlsanity.com/paper/2409.10197/) (2024)
- [Streaming Video Question-Answering with In-context Video KV-Cache Retrieval](https://mlsanity.com/paper/2503.00540/) (2025)
- [Baichuan-Omni-1.5 Technical Report](https://mlsanity.com/paper/2501.15368/) (2025)
- [Daily-Omni: Towards Audio-Visual Reasoning with Temporal Alignment across Modalities](https://mlsanity.com/paper/2505.17862/) (2025)
- [Mini-Omni2: Towards Open-source GPT-4o with Vision, Speech and Duplex Capabilities](https://mlsanity.com/paper/2410.11190/) (2024)
- [DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Models](https://mlsanity.com/paper/2411.15024/) (2024)
- [PruneVid: Visual Token Pruning for Efficient Video Large Language Models](https://mlsanity.com/paper/2412.16117/) (2024)
- [OmniVideoBench: Towards Audio-Visual Understanding Evaluation for Omni MLLMs](https://mlsanity.com/paper/2510.10689/) (2025)
- [HoliTom: Holistic Token Merging for Fast Video Large Language Models](https://mlsanity.com/paper/2505.21334/) (2025)
- [TopV: Compatible Token Pruning with Inference Time Optimization for Fast and Low-Memory Multimodal Vision Language Model](https://mlsanity.com/paper/2503.18278/) (2025)
- [LongVALE: Vision-Audio-Language-Event Benchmark Towards Time-Aware Omni-Modal Perception of Long Videos](https://mlsanity.com/paper/2411.19772/) (2024)
- [ZipVL: Efficient Large Vision-Language Models with Dynamic Token Sparsification](https://mlsanity.com/paper/2410.08584/) (2024)
- [InteractiveOmni: A Unified Omni-modal Model for Audio-Visual Multi-turn Dialogue](https://mlsanity.com/paper/2510.13747/) (2025)
- [AVoCaDO: An Audiovisual Video Captioner Driven by Temporal Orchestration](https://mlsanity.com/paper/2510.10395/) (2025)
- [OmniSIFT: Modality-Asymmetric Token Compression for Efficient Omni-modal Large Language Models](https://mlsanity.com/paper/2602.04804/) (2026)
- [From Specific-MLLMs to Omni-MLLMs: A Survey on MLLMs Aligned with Multi-modalities](https://mlsanity.com/paper/2412.11694/) (2024)
- [OmniZip: Audio-Guided Dynamic Token Compression for Fast Omnimodal Large Language Models](https://mlsanity.com/paper/2511.14582/) (2025)
- [OmniGAIA: Towards Native Omni-Modal AI Agents](https://mlsanity.com/paper/2602.22897/) (2026)
- [SpeechPrune: Context-aware Token Pruning for Speech Information Retrieval](https://mlsanity.com/paper/2412.12009/) (2024)

## Built on by (most cited first, 2 of 2)

- [Learning to Reason with Compressed Context: Ground-Truth-Free Adaptation of OmniLLMs via Self-Distillation](https://mlsanity.com/paper/2609.39953/) (2026)
- [LongEmo: Towards Emotion Understanding and Reasoning in Long Videos](https://mlsanity.com/paper/2609.40079/) (2026)

## Abstract (arXiv)

Omni modal large language models (OmniLLMs) have attracted wide attention for their ability to jointly process audio and video, but they generate large token sequences under audio-visual inputs, leading to substantial inference cost. Existing audio-visual token compression methods often rely on unimodal guidance, overlooking the temporal locality of query-relevant evidence in audio-visual inputs and implicitly assuming that the two modalities share a temporally aligned information density distribution. We propose OmniFocus, a training-free query-guided token compression method for OmniLLMs that performs independent importance estimation for video and audio, enabling a modality-symmetric compression design that preserves modality-specific salient evidence while maintaining audio-visual alignment, thereby mitigating the modality bias issue that can arise from unimodal-guided compression. Experiments on the Qwen2.5-Omni model family across four audio-visual benchmarks show that OmniFocus maintains strong compressed performance at low token retention ratios and outperforms existing baselines on several major benchmark scores at 25% token retention. On DailyOmni with Qwen2.5-Omni-7B at 25% token retention, OmniFocus maintains 59.40 accuracy while delivering up to 1.38× prefill speedup relative to the full-token baseline, highlighting a favorable practical accuracy-efficiency trade-off.