Kimi Linear
Hybrid linear-attention MoE with 48B total and 3B active parameters for long-context throughput.
About
Kimi Linear is Moonshot AI's hybrid linear-attention architecture, introduced in the Kimi Linear technical report (November 2025). Its core is Kimi Delta Attention (KDA), a gated linear-attention variant with per-channel forget gates, interleaved with full Multi-Head Latent Attention layers at a 3:1 ratio.
The flagship checkpoint is Kimi-Linear-48B-A3B-Instruct: a 48B-parameter MoE that activates only 3B parameters per token. The design cuts KV-cache usage by up to 75% and delivers up to 6x decoding throughput at 1M context versus full attention, making it a strong pick for long-context serving. It ships on Hugging Face under a modified MIT license, with the KDA kernel and vLLM implementations open-sourced alongside it.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 105; Trending score 80/100 from researcher assessment.