# Cross-Model KV Cache Transfer in LLM Families: A Closed-Form Linear Mapping for Prefill Reuse

arXiv: https://arxiv.org/abs/2608.03893 · HTML: https://mlsanity.com/paper/2608.03893/

## Key facts

- Cited by 15 papers in the map (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 26 of its 29 references that are in the map.
- First posted on arXiv 4 Aug 2026.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 26 of 26)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [CoQA: A Conversational Question Answering Challenge](https://mlsanity.com/paper/1808.07042/) (2018)
- [The Platonic Representation Hypothesis](https://mlsanity.com/paper/2405.07987/) (2024)
- [Prompt Cache: Modular Attention Reuse for Low-Latency Inference](https://mlsanity.com/paper/2311.04934/) (2023)
- [Reducing Transformer Key-Value Cache Size with Cross-Layer Attention](https://mlsanity.com/paper/2405.12981/) (2024)
- [NVIDIA Nemotron 3: Efficient and Open Intelligence](https://mlsanity.com/paper/2512.20856/) (2025)
- [Cache-to-Cache: Direct Semantic Communication Between Large Language Models](https://mlsanity.com/paper/2510.03215/) (2025)
- [Layer-Condensed KV Cache for Efficient Inference of Large Language Models](https://mlsanity.com/paper/2405.10637/) (2024)
- [Transferring Linear Features Across Language Models With Model Stitching](https://mlsanity.com/paper/2506.06609/) (2025)
- [SwiftKV: Fast Prefill-Optimized Inference with Knowledge-Preserving Model Transformation](https://mlsanity.com/paper/2410.03960/) (2024)
- [Latent Space Communication via K-V Cache Alignment](https://mlsanity.com/paper/2601.06123/) (2026)
- [Speculative Prefill: Turbocharging TTFT with Lightweight and Training-Free Token Importance Estimation](https://mlsanity.com/paper/2502.02789/) (2025)
- [Cross-model Transferability among Large Language Models on the Platonic Representations of Concepts](https://mlsanity.com/paper/2501.02009/) (2025)
- [IAM: Efficient Inference through Attention Mapping between Different-scale LLMs](https://mlsanity.com/paper/2507.11953/) (2025)
- [ICaRus: Identical Cache Reuse for Efficient Multi Model Inference](https://mlsanity.com/paper/2603.13281/) (2026)
- [Linear Representation Transferability Hypothesis: Leveraging Small Models to Steer Large Models](https://mlsanity.com/paper/2506.00653/) (2025)
- [PrefillShare: A Shared Prefill Module for KV Reuse in Multi-LLM Disaggregated Serving](https://mlsanity.com/paper/2602.12029/) (2026)
- [Cross-Family Speculative Prefill: Training-Free Long-Context Compression with Small Draft Models](https://mlsanity.com/paper/2603.02631/) (2026)

## Built on by (most cited first, 15 of 15)

- [CacheBridge: Efficient Cross-Model KV Cache Transfer](https://mlsanity.com/paper/2609.00891/) (2026)
- [A Universal Context-Reuse Layer for Cross-Model KV Sharing](https://mlsanity.com/paper/2608.30963/) (2026)
- [KV-Lingo: Learning KV-Cache Translators with Distillation](https://mlsanity.com/paper/2609.32610/) (2026)
- [Unified AI Gateway: A Framework for Joint Model Routing and KV Cache Management](https://mlsanity.com/paper/2609.06940/) (2026)
- [Shared-Prefix KV Reuse Across Standard LoRA Adapters: Quality and Serving Tradeoffs](https://mlsanity.com/paper/2609.17109/) (2026)
- [LatentPort: Beyond KV Cache - Cross-Model Transfer of Recurrent Memory in Hybrid Language Models: A 4B-to-9B Hybrid-State Handoff Without Target Prefix Replay](https://mlsanity.com/paper/2609.25053/) (2026)
- [Disaggregated Quantization: Specializing LLM Prefill and Decode](https://mlsanity.com/paper/2609.26333/) (2026)
- [Prefill-Free Cross-Family KV Cache Transfer for Heterogeneous Multi-Agent LLMs](https://mlsanity.com/paper/2609.32259/) (2026)
- [Positions Are Not Facts: The Mismatch Between KV Caches and Memory](https://mlsanity.com/paper/2609.33759/) (2026)
- [PReCache: Efficient KV Cache Sharing for Multi-LoRA Agents via Low-Rank Precomputation and Neutral Reconstruction](https://mlsanity.com/paper/2609.34054/) (2026)
- [KVCMAS: Efficient KV cache Correction for Shared Context in Multi-Agent Systems](https://mlsanity.com/paper/2609.34060/) (2026)
- [RaReCache: Bridging the Gap in Cross-Model KV Cache Reuse via Rank disagreement-based Selective Recomputation](https://mlsanity.com/paper/2610.11358/) (2026)
- [Swarm Telepathy: Prefill-Free KV-Cache Transfer Across Families](https://mlsanity.com/paper/2610.15095/) (2026)
- [CacheLink: Functional KV Transport for Zero-Prefill Cross-Model Context Handoff](https://mlsanity.com/paper/2610.15509/) (2026)
- [To Cache or Not to Cache: Understanding and Mitigating Safety Incompatibility in Cross-Model KV Cache Reuse](https://mlsanity.com/paper/2610.15721/) (2026)

## Abstract (arXiv)

Production deployments often swap between different-sized models in a family for cost-quality cascading, mid-conversation switching, and routing, and each swap forces the receiver to repay the prefill from scratch. We propose cross-model KV cache transfer, where the receiver reuses the source's KV cache, skipping prefill. We find that cross-model KV has substantial linear structure across matched-KV pairs, where source and target share KV head count and per-head dimension. On Qwen3 14B->32B, one source layer explains 56% of variance in the target's keys and 32% in values, rising to 79% and 65% with multiple source layers. Building on this, we design a closed-form ridge mapper that operates per head and proceeds in three steps. First, for each target layer we select the top-k most predictive source layers and concatenate their KV as input. Second, we strip RoPE from the keys before mapping, so the fit is position-free and reusable across context lengths. Third, we fit ridge regression on a small calibration set of 500 FineWeb-Edu sequences of 1,024 tokens each. Surprisingly, across six pairs in three families, this linear mapper retains 73-98% of the receiver's standalone-prefill accuracy on four pairs, while two degrade sharply. A nonlinear MLP recovers up to +37 pp HellaSwag retention on the failures. The mapper runs 2.7-25x faster than re-prefill and remains stable across multi-turn handoff, making cross-model KV cache transfer practical.