Loading
Life sciences · Preprint
arXiv · September 4, 2026
Early or partial results. Treat as a signal, not a conclusion.
KVMem is a GPU memory virtualization system for LLM agents that manages overflow of KV context to host and NVMe storage, reported to outperform compaction-based approaches on three long-context agent benchmarks. The work demonstrates feasibility on consumer hardware but remains unrefereed and lacks independent validation.
Empirical benchmark evaluation with comparative baseline. LLM agent workloads with persistent context histories; benchmarked using Qwen 3.6/3.8-27B models in local deployment setting.. Intervention: KVMem: KV-context virtualization system using paged KV state across GPU memory, host memory, and NVMe with lightweight attention-space indexes for historical block selection.. Compared with: Compaction-based context management (de facto standard for handling context overflow)..
DeepSWE benchmark: task success improved from 43.8% with compaction-only to 48.4% with KVMem on Qwen3.8-27B KVMem virtualizes agent workspaces up to 1M tokens on a 24 GB RTX 5090 Laptop GPU, four times the model's native 256K-token context window Single-session inference generates approximately 50 tokens/s, described as providing interactive responsiveness
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
The source did not state who this applies to in practice.
A systems engineering contribution presenting a novel technical approach with empirical evaluation on benchmarks, but lacking peer review, clinical or established ground-truth validation, and presented as a preprint without independent replication.
As stated by the source record.
Quoted from the source exactly as published.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
Modern LLM agents operate in persistent workspaces whose accumulated history can exceed both GPU KV capacity and the model's native context window. Existing systems typically compact older context into summaries or retrieve it later as text, either losing fine-grained execution evidence or repeatedly prefilling content that the model has already processed. We present KVMem, a KV-context virtualization system that preserves overflowed workspace history as paged KV state across GPU memory, host memory, and NVMe. KVMem uses lightweight, model-native attention-space indexes to select relevant historical blocks and materializes a query-dependent execution view bounded by the model's native context window. Extensive evaluations on long-context agent benchmarks spanning histories up to one million tokens, including LongMemEval, MemoryAgentBench, and AgentLongBench, show that KVMem generally achieves higher task utility and greater inference efficiency than compaction-based approaches, the de facto standard for handling context overflow. In the DeepSWE long-context test with Qwen3.8-27B, KVMem improves task success from 43.8% with compaction-only context management to 48.4%. In our local-deployment evaluation, KVMem runs Qwen3.6/3.8-27B NVFP4 with MTP on an off-the-shelf laptop equipped with a 24\,GB RTX 5090 Laptop GPU, virtualizing agent workspaces of up to 1M tokens-four times the model's native 256K-token context window. In a single-session setting, KVMem generates $\sim$50 tokens/s, providing interactive responsiveness for local agent execution. More broadly, by decoupling addressable workspace size from the LLM's native context window, KVMem provides a practical path toward long-running agents whose workspaces can grow beyond that window.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.