DeepSeek-V4.1-Flash: Pushing the Limits of KV…

DeepSeek-V4.1-Flash asks how a million-token agent can keep attention memory from overwhelming accelerator memory, persistent storage, and transfer bandwidth. Read it to see how a causal encoder-decoder shares compressed global context across depth, how three CSA2 layer modes divide cache and indexing work, and why bounded replay makes an explicit approximation to avoid persisting local attention state.

Reading focus: How a 20-layer causal encoder produces shared global context for a 20-layer decoder, reducing prompt-side activation from 16B to 8B parameters per token. How CSA2 Full, Reindex, and Reuse modes share global KV and sparse selections while preserving layer-local queries and sliding-window state. Why 890-byte global KV and one-eighth persistent cache are separate claims, and where approximate bounded replay can fail.

Technical Report 2026. DeepSeek-AI. 30 min read, medium difficulty.