SnapKV

LLM knows what you're seeking before generation. NeurIPS 2024.

At a glance

Type
KV-cache pre-selection
Tier
T3
Created
2024-04
Latest release
not applicable — not OSS
License
not applicable — not OSS
GitHub
not applicable — no GitHub repo
Pricing
not applicable — not commercial
Funding
not applicable — not commercial

Taxonomy

storage
kv-cache
retrieval
attention
persistence
session
update
evict-oldest
unit
kv-token
governance
n/a
conflict
n/a

When to use

Optimised for: not applicable - research paper

Anti-fit: not applicable - research paper

Pros & cons

Pros

LLM identifies what's relevant before generation, pre-selecting KV-cache to keep; NeurIPS 2024.

Cons

Pre-selection quality bounded by attention-pattern interpretability; production deployment limited.

Claims & capabilities

Fine-tuning-free KV cache compression identifying that each attention head focuses on specific prompt features; uses observation window at end of prompt to auto-select important KV positions; 3.6x decoding speed and 8.2x memory footprint improvement on 16K-token inputs; processes up to 380K context tokens on single A100-80GB; negligible accuracy drop on Needle-in-a-Haystack

Technical surface

API surface
not applicable — research paper
Backend storage
not applicable — research paper
Deployment
not applicable — not a deployable product
Embedding model
not applicable — research paper
Multi-tenancy
not applicable — research paper
MCP
not applicable — research paper, no deployed product
A2A
not applicable — research paper, no deployed product
OpenTelemetry
not applicable — research paper, no deployed product

Similar systems

Other recent method papers — theorized, no distinct product in the catalog, ranked by inbound references.

  • Compressive Transformer T3

    Maintains recent states in full resolution while compressing older memories with learned compression functions. DeepMind.

  • MemGPT v2 / agent-tools T3

    Already in catalog as the foundational MemGPT paper. Note: Letta is the productionised successor (cross-listed).

  • Transformer-XL T3

    Extends context through segment-level recurrence + caching of hidden states from prior segments. Foundational long-context architecture.

  • Generative Agents T3

    Park et al. — landmark agent-simulation paper. Reflection + memory stream + retrieval enable believable agent behavior.

  • MemoryBank T3

    Enhances LLMs with long-term memory. Early influential paper.

  • Reflexion T3

    Language agents with verbal reinforcement learning.

Related systems

References (1)

  • LongBench cites — S2 isInfluential citation

Row last verified 2026-05-14. Catalog data is CC-BY-4.0 — see how to read this.