GRKV (KV-cache compression): current frontier — recent, not yet superseded in the knowledge base. 0 paper(s) critique it, 0 beat it on benchmarks — not ranked as a superseded baseline. Sub-problem: cluster led by SnapKV. Newer alternatives in the same sub-problem include STaR-KV, GRKV, MomentKV, NestedKV, IndexMem.

KV-cache compression

GRKV

GRKV: Global Regression for Training-Free KV Cache Compression in Long-Context LLMs

On the frontierfirst seen May 29, 2026

Current frontier — recent, not yet superseded in the knowledge base

0 papers critique it · 0 beat it on benchmarks

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.

STaR-KVSTaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models
Jun 1, 2026
GRKVGRKV: Global Regression for Training-Free KV Cache Compression in Long-Context LLMs
May 29, 2026
MomentKVMoment-KV: Momentum-Based Decode-Time KV Cache Compression for Long Generation
May 28, 2026
NestedKVNestedKV: Nested Memory Routing for Long-Context KV Cache Compression
May 26, 2026
IndexMemIndexMem: Learned KV-Cache Eviction with Latent Memory for Long-Context LLM Inference
May 25, 2026
CONF-KVCONF-KV: Confidence-Aware KV Cache Eviction with Mixed-Precision Storage for Long-Horizon LLM
May 24, 2026
Meta-SoftMeta-Soft: Leveraging Composable Meta-Tokens for Context-Preserving KV Cache Compression
May 21, 2026
KV-FoldKV-Fold: One-Step KV-Cache Recurrence for Long-Context Inference
May 12, 2026
Global Retention-Based KV EvictionMake Each Token Count: Towards Improving Long-Context Performance with KV Cache Eviction
May 10, 2026
ReST-KVReST-KV: Robust KV Cache Eviction with Layer-wise Output Reconstruction and Spatial-Temporal Smoothing
May 9, 2026
LaProxReformulating KV Cache Eviction Problem for Long-Context LLM Inference
May 8, 2026