Transformer Architecture · 2025

Kimi Linear: An Expressive, Efficient Attention Architecture

Kimi Team (Moonshot AI)

Kimi Delta Attention plus an MLA hybrid is the first linear-attention architecture to beat full attention across short-context, long-context, and RL regimes while cutting KV cache and speeding decode.

Editorial record

Plain-language summary

Kimi Delta Attention plus an MLA hybrid is the first linear-attention architecture to beat full attention across short-context, long-context, and RL regimes while cutting KV cache and speeding decode.

Knowledge graph

Relationships

Descendants

  • Depends onEvidence: Strongly supported

    Attention Is All You Need

    Builds on the transformer architecture lineage in the archive

    freshness sweep 2026

Source record

Provenance

Record ID
P-618
Record created
2026-07-13
Last reviewed
2026-07-14
Record version
2

Citation caveat: Citation metadata is approximate and marked unverified in the source dataset.