# KV cache quantization

Area: [Efficient LLM inference and serving](https://mlsanity.com/inference/index.md). HTML: https://mlsanity.com/inference/kv-cache-quantization/

## Key facts

- KV cache quantization: 122 papers on arXiv since 2023 (data through 7 Oct 2026).
- 75 of its 122 mapped papers (61%) cite “KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization” (2024, arXiv 2401.18079).
- 74 of its 122 mapped papers (61%) cite “KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache” (2024, arXiv 2402.02750).
- 8 new papers 24 Sep–7 Oct.

## Most cited here

- [KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization](https://mlsanity.com/paper/2401.18079/) (2024), cited by 75 papers here
- [KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache](https://mlsanity.com/paper/2402.02750/) (2024), cited by 74 papers here
- [QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs](https://mlsanity.com/paper/2404.00456/) (2024), cited by 51 papers here
- [SpinQuant: LLM quantization with learned rotations](https://mlsanity.com/paper/2405.16406/) (2024), cited by 30 papers here
- [Atom: Low-bit Quantization for Efficient and Accurate LLM Serving](https://mlsanity.com/paper/2310.19102/) (2023), cited by 29 papers here
- [QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving](https://mlsanity.com/paper/2405.04532/) (2024), cited by 29 papers here
- [ZipCache: Accurate and Efficient KV Cache Quantization with Salient Token Identification](https://mlsanity.com/paper/2405.14256/) (2024), cited by 20 papers here
- [QAQ: Quality Adaptive Quantization for LLM KV Cache](https://mlsanity.com/paper/2403.04643/) (2024), cited by 19 papers here
- [KV Cache is 1 Bit Per Channel: Efficient Large Language Model Inference with Coupled Quantization](https://mlsanity.com/paper/2405.03917/) (2024), cited by 16 papers here
- [QJL: 1-Bit Quantized JL Transform for KV Cache Quantization with Zero Overhead](https://mlsanity.com/paper/2406.03482/) (2024), cited by 16 papers here

## New (24 Sep–7 Oct)

- [ResidualQuant: KV Cache Quantization for Looped Transformers with 2-Bit Residuals](https://mlsanity.com/paper/2610.10381/) (2026-10-07)
- [Dual-QK: Sharp Queries and Flat Keys for Prunable 2-bit KV Caches](https://mlsanity.com/paper/2610.09827/) (2026-10-07)
- [Few Bits, One Law: Toward W2A4KV2](https://mlsanity.com/paper/2610.09202/) (2026-10-06)
- [AlignQuant: Tile-Aligned Mixed-Precision Quantization for Efficient LLM Generation](https://mlsanity.com/paper/2610.07457/) (2026-10-05)
- [Tailoring the Quantization Space for 1-Bit KV Cache Compression](https://mlsanity.com/paper/2610.03027/) (2026-10-02)
- [STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization](https://mlsanity.com/paper/2609.38169/) (2026-09-29)
- [WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms](https://mlsanity.com/paper/2609.38121/) (2026-09-29)
- [Behavioral Capacity Certificates for Quantized Language Models](https://mlsanity.com/paper/2609.37887/) (2026-09-29)