paper-with-me

Papers

Cache-Aware Prompt Compression:A Two-Tier Cost Model for LLM API Caching

2026-07-17 · Yan Song arxiv

Production LLM deployments combine two cost-reduction primitives: prompt caching (a discounted rate for re-used token prefixes) and prompt compression (fewer tokens sent). The compression literature has standardized on query-aware methods that produce a different compressed prefix per query, mechanically invalidating the prefix-strict cache on every call. We characterize this cost empirically on Anthropic's Sonnet 4.6 API and find caching is far from the rho=1.0 ideal the literature assumes: Sonnet's cache has a two-tier architecture with a sharp threshold near 3,500 tokens, below which the hit rate plateaus at rho~0.83 across 30-call sessions. Our cost model predicts, and experiments confirm, that under realistic rho, query-aware compression beats naive caching at high compression ratios (r>=6). We propose Cache-Aware Prompt Compression (CAPC), pairing query-agnostic compression with explicit cache_control plus a tier-preserving ratio bound that prevents over-compression from pushing the cached prefix into the hot tier. CAPC is the cheapest strategy in 16/16 configurations on LongBench-v2, with mean savings of 49% over cache-only, 64% over query-aware compression, and 90% over vanilla, at quality within 0.05 of the uncompressed baseline. We validate CAPC on three production workloads: an enterprise tool-using assistant with a 94k-token schema prefix (51.7% cost reduction at r=3); a graphify knowledge-graph RAG pipeline across two codebases (9.3x vs cache-all on FastAPI, 2.4x on httpx); and the public tau-bench retail benchmark (50 tasks), where CAPC is the cheapest of four strategies with reward exactly equal to vanilla (both 36/50, p=1.00) while query-aware compression is the most expensive at +40.1% over vanilla -- the first production confirmation of the crossover model's negative-ROI prediction on a public benchmark.

📄 PDF Abstract BibTeX arXiv:2607.15516

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

S$^4$R: Selective Sampling, Subspaces, and Sparse Reconstruction for Compressed Long-Context KV Caching

2026-08-01 · Jialong Han, You Wu, Kewei Tu arxiv

The growth of context window lengths in Large Language Models (LLMs) significantly enhances their long-context capabilities but incurs prohibitive memory costs due to the Key-Value (KV) cache. Although low-rank compressi…

PANDO: Efficient Multimodal AI Agents via Online Skill Distillation

2026-05-24 · Yubo Li, Yidi Miao, Yuntian Shen, Yuxin Liu arxiv

Recent advances in multimodal web agents often rely on increased inference-time computation, including rollout search, verifier passes, offline skill discovery, and specialist model stacks. This raises a central question…

EVICPRESS: Joint KV-Cache Compression and Eviction for Efficient LLM Serving

2025-12-16 · Shaoting Feng, Yuhan Liu, Hanchen Li, Xiaokun Chen 외 arxiv

Reusing KV cache is essential for high efficiency of Large Language Model (LLM) inference systems. With more LLM users, the KV cache footprint can easily exceed GPU memory capacity, so prior work has proposed to either e…

The Pitfalls of KV Cache Compression

2025-09-30 · Alex Chen, Renato Geh, Aditya Grover, Guy Van den Broeck 외 arxiv

KV cache compression promises increased throughput and efficiency with negligible loss in performance. While the gains in throughput are indisputable and recent literature has indeed shown minimal degradation on particul…

AnchorKV: Safety-Aware KV Cache Compression via Soft Penalty with a Refusal Anchor

2026-06-16 · Ning Ni, Yingjie Lao arxiv

Large language models (LLMs) outperform earlier architectures on generative inference and long-context tasks, but their large size introduces significant challenges in memory usage, energy cost, and on-device deployment.…