# Multi-Granular Spatio-Temporal Token Merging for Training-Free Acceleration of Video LLMs

arXiv: https://arxiv.org/abs/2507.07990 · HTML: https://mlsanity.com/paper/2507.07990/

## Key facts

- Cited by 31 papers in the map (data through 14 Oct 2026).
- Placed in Visual token pruning in multimodal models, one of 12 directions in Efficient LLM inference and serving.
- Builds on 48 of its 62 references that are in the map.
- First posted on arXiv 10 Jul 2025.

## Placed in

- Efficient LLM inference and serving / [Visual token pruning in multimodal models](https://mlsanity.com/inference/visual-token-pruning-in-multimodal-models/)

## Builds on (most cited first, 30 of 48)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset](https://mlsanity.com/paper/1705.07750/) (2017)
- [Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context](https://mlsanity.com/paper/2403.05530/) (2024)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding](https://mlsanity.com/paper/2306.02858/) (2023)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [Video-LLaVA: Learning United Visual Representation by Alignment Before Projection](https://mlsanity.com/paper/2311.10122/) (2023)
- [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://mlsanity.com/paper/2203.12602/) (2022)
- [MVBench: A Comprehensive Multi-modal Video Understanding Benchmark](https://mlsanity.com/paper/2311.17005/) (2023)
- [Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models](https://mlsanity.com/paper/2306.05424/) (2023)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [VideoChat: Chat-Centric Video Understanding](https://mlsanity.com/paper/2305.06355/) (2023)
- [NExT-QA:Next Phase of Question-Answering to Explaining Temporal Actions](https://mlsanity.com/paper/2105.08276/) (2021)
- [EgoSchema: A Diagnostic Benchmark for Very Long-form Video Language Understanding](https://mlsanity.com/paper/2308.09126/) (2023)
- [LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding](https://mlsanity.com/paper/2407.15754/) (2024)
- [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)
- [MovieChat: From Dense Token to Sparse Memory for Long Video Understanding](https://mlsanity.com/paper/2307.16449/) (2023)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [Long Context Transfer from Language to Vision](https://mlsanity.com/paper/2406.16852/) (2024)

## Built on by (most cited first, 30 of 31)

- [A Survey of Token Compression for Efficient Multimodal Large Language Models](https://mlsanity.com/paper/2507.20198/) (2025)
- [FlashVID: Efficient Video Large Language Models via Training-free Tree-based Spatiotemporal Token Merging](https://mlsanity.com/paper/2602.08024/) (2026)
- [VideoNSA: Native Sparse Attention Scales Video Understanding](https://mlsanity.com/paper/2510.02295/) (2025)
- [HTTM: Head-wise Temporal Token Merging for Faster VGGT](https://mlsanity.com/paper/2511.21317/) (2025)
- [Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospects](https://mlsanity.com/paper/2604.05546/) (2026)
- [VideoTemp-o3: Harmonizing Temporal Grounding and Video Understanding in Agentic Thinking-with-Videos](https://mlsanity.com/paper/2602.07801/) (2026)
- [FLoC: Facility Location-Based Efficient Visual Token Compression for Long Video Understanding](https://mlsanity.com/paper/2511.00141/) (2025)
- [EchoingPixels: Aliasing-Resistant Joint Token Reduction for Audio-Visual LLMs](https://mlsanity.com/paper/2512.10324/) (2025)
- [D3ToM: Decider-Guided Dynamic Token Merging for Accelerating Diffusion MLLMs](https://mlsanity.com/paper/2511.12280/) (2025)
- [Stateful Token Reduction for Long-Video Hybrid VLMs](https://mlsanity.com/paper/2603.00198/) (2026)
- [ForestPrune: High-ratio Visual Token Compression for Video Multimodal Large Language Models via Spatial-Temporal Forest Modeling](https://mlsanity.com/paper/2603.22911/) (2026)
- [TIR-Flow: Active Video Search and Reasoning with Frozen VLMs](https://mlsanity.com/paper/2601.06176/) (2026)
- [Keep What Audio Cannot Say: Context-Preserving Token Pruning for Omni-LLMs](https://mlsanity.com/paper/2605.11605/) (2026)
- [OmniRefine: Alignment-Aware Cooperative Compression for Efficient Omnimodal Large Language Models](https://mlsanity.com/paper/2605.12056/) (2026)
- [Video Parallel Scaling: Aggregating Diverse Frame Subsets for VideoLLMs](https://mlsanity.com/paper/2509.08016/) (2025)
- [Stay Unique, Stay Efficient: Preserving Model Personality in Multi-Task Merging](https://mlsanity.com/paper/2512.01461/) (2025)
- [DynaTok: Temporally Adaptive and Positional Bias-Aware Token Compression for Video-LLMs](https://mlsanity.com/paper/2605.19322/) (2026)
- [ReGATE: Learning Faster and Better with Fewer Tokens in MLLMs](https://mlsanity.com/paper/2507.21420/) (2025)
- [Foundry: Distilling 3D Foundation Models for the Edge](https://mlsanity.com/paper/2511.20721/) (2025)
- [Unified Spatio-Temporal Token Scoring for Efficient Video VLMs](https://mlsanity.com/paper/2603.18004/) (2026)
- [OTT-Vid: Optimal Transport Temporal Token Compression for Video Large Language Models](https://mlsanity.com/paper/2605.11803/) (2026)
- [Sharp Eyes and Memory for VideoLLMs: Information-Aware Visual Token Pruning for Efficient and Reliable VideoLLM Reasoning](https://mlsanity.com/paper/2511.08003/) (2025)
- [ShaRP: SHAllow-LayeR Pruning for Efficient Video Large Language Models](https://mlsanity.com/paper/2512.05385/) (2025)
- [Towards Efficient Large Vision-Language Models: A Comprehensive Survey on Inference Strategies](https://mlsanity.com/paper/2603.27960/) (2026)
- [Sink-Token-Aware Pruning for Fine-Grained Video Understanding in Efficient Video LLMs](https://mlsanity.com/paper/2604.20937/) (2026)
- [VLMaxxing through FrameMogging Training-Free Anti-Recomputation for Video Vision-Language Models](https://mlsanity.com/paper/2605.03351/) (2026)
- [Zero-Shot 3D Question Answering via Hierarchical View-to-Token Transportation](https://mlsanity.com/paper/2606.03100/) (2026)
- [CRAFT: Compression via Recursive Adaptive Fusion of Video Tokens for Vision-Language Models](https://mlsanity.com/paper/2608.01644/) (2026)
- [CARVE: Cross-Slice Anisotropic Reallocation of Visual Evidence for Efficient 3D Medical Volume Understanding](https://mlsanity.com/paper/2608.04515/) (2026)
- [Why Is Video Still So Expensive? A Survey of Inference-Efficiency Mechanisms in Video and Audiovisual LLMs](https://mlsanity.com/paper/2609.10355/) (2026)

## Abstract (arXiv)

Video large language models (LLMs) achieve strong video understanding by leveraging a large number of spatio-temporal tokens, but suffer from quadratic computational scaling with token count. To address this, we propose a training-free spatio-temporal token merging method, named STTM. Our key insight is to exploit local spatial and temporal redundancy in video data which has been overlooked in prior work. STTM first transforms each frame into multi-granular spatial tokens using a coarse-to-fine search over a quadtree structure, then performs directed pairwise merging across the temporal dimension. This decomposed merging approach outperforms existing token reduction methods across six video QA benchmarks. Notably, STTM achieves a 2× speed-up with only a 0.5% accuracy drop under a 50% token budget, and a 3× speed-up with just a 2% drop under a 30% budget. Moreover, STTM is query-agnostic, allowing KV cache reuse across different questions for the same video. The project page is available at https://www.jshyun.me/projects/sttm.