Architecture atlas

Model Architectures

Interactive architecture references with clickable components and tensor contracts.

Hybrid recurrent architecture

Kimi K3

Explore Kimi K3 from token flow down to MLA, short convolution, gated delta-rule state updates, MoE routing, and kernel tensor shapes.

  • MLA
  • Delta rule
  • MoE
  • Kernel internals
Explore architecture →

Million-token architecture

DeepSeek V4

Trace DeepSeek V4 through CSA and HCA, the lightning indexer, manifold-constrained residuals, DeepSeekMoE, and multi-token prediction.

  • CSA / HCA
  • mHC
  • DeepSeekMoE
  • MTP
Explore architecture →