mlsanityData to 14 Oct 2026

R-KV: Redundancy-aware KV Cache Compression for Reasoning Models

arXiv 2505.24133 ↗First posted 30 May 2025cs.CL cs.AIcited by 54

On the map: LLM reasoning → Test-time compute and efficient thinking

also in 1 more
Abstract (arXiv)

Reasoning models have demonstrated impressive performance in self-reflection and chain-of-thought reasoning. However, they often produce excessively long outputs, leading to prohibitively large key-value (KV) caches during inference. While chain-of-thought inference significantly improves performance on complex reasoning tasks, it can also lead to reasoning failures when deployed with existing KV cache compression approaches. To address this, we propose Redundancy-aware KV Cache Compression for Reasoning models (R-KV), a novel method specifically targeting redundant tokens in reasoning models. Our method preserves nearly 100% of the full KV cache performance using only 10% of the KV cache, substantially outperforming existing KV cache baselines, which reach only 60% of the performance. Remarkably, R-KV even achieves 105% of full KV cache performance with 16% of the KV cache. This KV-cache reduction also leads to a 90% memory saving and a 6.6X throughput over standard chain-of-thought reasoning inference. Experimental results show that R-KV consistently outperforms existing KV cache compression baselines across two mathematical reasoning datasets.

Key facts. Cited by 54 papers in the map (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 27 of its 32 references that are in the map. First posted on arXiv 30 May 2025.

Builds on

27 of 32 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025Thought Manipulation: External Thought Can Be Efficient for Large Reasoning Models18 Apr
  2. 2025Reasoning Models Can Be Effective Without Thinking14 Apr
  3. 2025Concise Reasoning via Reinforcement Learning7 Apr
  4. 2024Token-Budget-Aware LLM Reasoning24 Dec
  5. 2024Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning25 Oct
  6. 2024Eigen Attention: Attention in Low-Rank Space for KV Cache Compression10 Aug

Other references · 21

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  3. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  4. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  5. 2023Efficient Streaming Language Models with Attention Sinkscited by 2,186
  6. 2024Qwen2 Technical Reportcited by 2,120
  7. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,210
  8. 2024SnapKV: LLM Knows What You are Looking for Before Generationcited by 880
  9. 2023H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Modelscited by 832
  10. 2024KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cachecited by 637
  11. 2023Model Tells You What to Discard: Adaptive KV Cache Compression for LLMscited by 474
  12. 2024PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funnelingcited by 432
9 more
  1. 2024Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inferencecited by 235
  2. 2024ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inferencecited by 134
  3. 2024Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inferencecited by 116
  4. 2024WKVQuant: Quantizing Weight and Key/Value Cache for Large Language Models Gains Morecited by 69
  5. 2024LoRC: Low-Rank Compression for LLMs KV Cache with a Progressive Compression Strategycited by 51
  6. 2025Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Modelscited by 51
  7. 2023Compressed Context Memory For Online Language Model Interactioncited by 43
  8. 2025Think When You Need: Self-Adaptive Chain-of-Thought Learningcited by 38
  9. 2025Adaptive Group Policy Optimization: Towards Stable Training and Token-Efficient Reasoningcited by 21

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 7; 2026: 47.

54 papers.

Sort:

By area:

  1. 2025SeerAttention-R: Sparse Attention Adaptation for Long Reasoningcited by 30
  2. 2025ThinKV: Thought-Adaptive KV Cache Compression for Efficient Reasoning Modelscited by 21
  3. 2026TriAttention: Efficient Long Reasoning with Trigonometric KV Compressioncited by 21
  4. 2026MEMENTO: Teaching LLMs to Manage Their Own Contextcited by 18
  5. 2026WorldKV: Efficient World Memory with World Retrieval and Compressioncited by 15
  6. 2026SideQuest: Model-Driven KV Cache Management for Long-Horizon Agentic Reasoningcited by 11
  7. 2025Which Heads Matter for Reasoning? RL-Guided KV Cache Compressioncited by 10
  8. 2025KaVa: Latent Reasoning via Compressed KV-Cache Distillationcited by 8
  9. 2026Value-Aware Stochastic KV Cache Eviction for Reasoning Modelscited by 8
  10. 2025DELTA: Dynamic Layer-Aware Token Attention for Efficient Long-Context Reasoningcited by 6
  11. 2026LongFlow: Efficient KV Cache Compression for Reasoning Modelscited by 6
  12. 2026FROST: Filtering Reasoning Outliers with Attention for Efficient Reasoningcited by 5
38 more
  1. 2026When Does Value-Aware KV Eviction Help? A Fixed-Contract Diagnostic for Non-Monotone Cache Compressioncited by 5
  2. 2025Not All Bits Are Equal: Scale-Dependent Memory Optimization Strategies for Reasoning Modelscited by 4
  3. 2025KV Admission: Learning What to Write for Efficient Long-Context LLM Inferencecited by 4
  4. 2026DashAttention: Differentiable and Adaptive Sparse Hierarchical Attentioncited by 4
  5. 2026MoE-nD: Per-Layer Mixture-of-Experts Routing for Multi-Axis KV Cache Compressioncited by 2
  6. 2026Epiphany-Aware KV Cache Eviction Without the Attention Matrixcited by 2
  7. 2026KARA: Efficient Reasoning LLM Serving via Sliding-Window KV Cache Compressioncited by 2
  8. 2026Does Accuracy Equal Evidence? Reasoning Faithfulness under KV Cache Compressioncited by 2
  9. 2026KV-Rescue: Recovering Reasoning Language Model KV Eviction Loss via Stepwise Interleavingcited by 2
  10. 2026Prefix Sliding for efficient test-time scalingcited by 2
  11. 2026SAS: Simple Attention Sparsification via End-to-End Optimization of Context Rankingcited by 2
  12. 2026Transactional Attention: Semantic Sponsorship for KV-Cache Retentioncited by 1
  13. 2026CacheFlow: Efficient LLM Serving with 3D-Parallel KV Cache Restorationcited by 1
  14. 2026AQuaUI: Visual Token Reduction for GUI Agents with Adaptive Quadtreescited by 1
  15. 2026Adaptive Mass-Segmented KV Compression for Long-Context Reasoningcited by 1
  16. 2026DepthWeave-KV: Token-Adaptive Cross-Layer Residual Factorization for Long-Context KV Cache Compressioncited by 1
  17. 2026LOCKS: Page-Local Compact Key Summaries for Efficient Long-Context Decodingcited by 1
  18. 2026Random Attention: Rethinking KV Cache Eviction for Efficient Reasoningcited by 1
  19. 2026EchoPress: Query-Agnostic KV Cache Pruning via Virtual Context Reconstructioncited by 1
  20. 2026Behavior-Preserving KV Cache Compressioncited by 1
  21. 2026SageSched: Efficient LLM Scheduling Confronting Demand Uncertainty and Hybriditycited by 0
  22. 2026CASK: Core-Aware Selective KV Compression for Reasoning Tracescited by 0
  23. 2026Salca: A Sparsity-Aware Hardware Accelerator for Efficient Long-Context Attention Decodingcited by 0
  24. 2026Minimal-Intervention KV Retention via Set-Conditioned Diversitycited by 0
  25. 2026FreqDepthKV: Frequency-Guided Depth Sharing for Robust KV Cache Compression in Long-Context LLM Inferencecited by 0
  26. 2026CommitKV: Lifecycle-Aware KV Cache Compression via Commit Transitions for Multi-Turn Agentscited by 0
  27. 2026Batch-wise Adaptive Pruning: Periodic Neuron Activation-Aware Weight Pruning for Language Reasoning Modelcited by 0
  28. 2026Small Frequency Corrections Can Change What Survives KV Cache Compressioncited by 0
  29. 2026GrowPage: On-Demand KV Budgeting for Efficient LLM Reasoning Servingcited by 0
  30. 2026ECOKV: Geometry-Aware KV Cache Eviction via Complementary Diversity Metricscited by 0
  31. 2026AgentKV: Phase-Aware KV Eviction for Agentic LLMscited by 0
  32. 2026StepKV: Step-Aware KV Cache Compression for LLM Agentscited by 0
  33. 2026Beyond Mean Attention: Diversity-Aware, Layer-Wise Scoring for KV Cache Evictioncited by 0
  34. 2026SparseEngine: Sparse-First Inference Enginecited by 0
  35. 2026CORE: COverage CAlibration and Evicted-Mass REdistribution for KV Cachecited by 0
  36. 2026WakeKV: Reactive, Reversible KV Residency for Heads That Change Their Mindscited by 0
  37. 2026iS-KV: Online Low-Rank KV Cache Compression via Block-Incremental SVDcited by 0
  38. 2026AvoKV-E: Payload-Aware KV Cache Eviction for Long Reasoningcited by 0

50 most cited of 54 All 54 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025SkipKV: Selective Skipping of KV Generation and Storage for Efficient Inference with Large Reasoning Models
  2. 2025Semantic Integrity Matters: Benchmarking and Preserving High-Density Reasoning in KV Cache Compression
  3. 2025Hold Onto That Thought: Assessing KV Cache Compression On Reasoning
  4. 2026BeaconKV: Key-Value Cache Compression Guided by Beacon Queries for Efficient Large Reasoning Model Inference
  5. 2026Fewer Tokens, Smaller Cache: Reward-Coordinated Efficient Reasoning
  6. 2025Breadcrumbs Reasoning: Memory-Efficient Reasoning with Compression Beacons
  7. 2025Reasoning Path Compression: Compressing Generation Trajectories for Efficient LLM Reasoning
  8. 2025Lethe: Layer- and Time-Adaptive KV Cache Pruning for Reasoning-Intensive LLM Serving