# CacheBridge: Efficient Cross-Model KV Cache Transfer

arXiv: https://arxiv.org/abs/2609.00891 · HTML: https://mlsanity.com/paper/2609.00891/

## Key facts

- Cited by 6 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 19 of its 20 references that are in the map.
- First posted on arXiv 1 Sep 2026.

## Placed in

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

## Builds on (most cited first, 19 of 19)

- [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)
- [AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation](https://mlsanity.com/paper/2308.08155/) (2023)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale](https://mlsanity.com/paper/2406.17557/) (2024)
- [FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance](https://mlsanity.com/paper/2305.05176/) (2023)
- [RouteLLM: Learning to Route LLMs with Preference Data](https://mlsanity.com/paper/2406.18665/) (2024)
- [Cache-to-Cache: Direct Semantic Communication Between Large Language Models](https://mlsanity.com/paper/2510.03215/) (2025)
- [Latent Space Communication via K-V Cache Alignment](https://mlsanity.com/paper/2601.06123/) (2026)
- [Cross-Model KV Cache Transfer in LLM Families: A Closed-Form Linear Mapping for Prefill Reuse](https://mlsanity.com/paper/2608.03893/) (2026)
- [IAM: Efficient Inference through Attention Mapping between Different-scale LLMs](https://mlsanity.com/paper/2507.11953/) (2025)
- [Efficient Multi-Adapter LLM Serving via Cross-Model KV-Cache Reuse with Activated LoRA](https://mlsanity.com/paper/2512.17910/) (2025)
- [ICaRus: Identical Cache Reuse for Efficient Multi Model Inference](https://mlsanity.com/paper/2603.13281/) (2026)
- [Latent Cache Flow: Model-to-Model Communication Without Text](https://mlsanity.com/paper/2605.22863/) (2026)
- [Fathom-DeepResearch: Unlocking Long Horizon Information Retrieval and Synthesis for SLMs](https://mlsanity.com/paper/2509.24107/) (2025)
- [Semantic Cache Distillation: Efficient State Transfer via Reuse and Selective Patching](https://mlsanity.com/paper/2606.07684/) (2026)
- [KV Cache Translation across Heterogeneous Large Language Models](https://mlsanity.com/paper/2607.28979/) (2026)
- [SmartGen: Seamless Disaggregated LLM Inference with Selective KV Cache Transfer](https://mlsanity.com/paper/2607.28150/) (2026)
- [PRISM: Fast Online LLM Serving via Scheduling-Memory Co-design](https://mlsanity.com/paper/2605.08581/) (2026)

## Built on by (most cited first, 6 of 6)

- [KV-Lingo: Learning KV-Cache Translators with Distillation](https://mlsanity.com/paper/2609.32610/) (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)
- [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)

Sharing context between LLMs in a multi-model system requires the receiving model to prefill the shared prefix because KV caches are model-specific. Recent closed-form cross-model KV transfer, hereafter Full-Head Mapping, avoids this replay by fitting a training-free affine mapper from source to target caches. However, its full-head design maps each target KV head from every source KV head in the selected layers, making transfer quality sensitive to architectural differences and causing mapper storage and application cost to grow with layer support. To this end, we introduce CacheBridge, which co-designs architecture-indexed mapper support, attention-aligned calibration, and bounded mapper construction while retaining a closed-form affine interface for online deployment. CacheBridge restricts each target head to a matched source head, weights reconstruction errors by causal attention sensitivity, and uses a fused GPU kernel to construct weighted sufficient statistics without materializing full observation tensors. Across three transfer directions, CacheBridge recovers the two Ministral 3 transfer directions where Full-Head Mapping loses substantial accuracy while preserving 99.83% mean target retention on Qwen3. On Qwen3 14B→32B, it reduces mapper storage by 8×, accelerates application by up to 3.0×, matches fullhead with one tenth of the calibration data, and reduces 500-sequence construction from 92.63 to 8.63 seconds (10.7×).