Unverified50% confidenceFactExact time
DeepSeek V4.1's SWA Bound replay technique reduces the number of Tokens that need to be recomputed when reading the question from 2560 to 128, roughly halving the total computation for reading the question.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/21/2026
First Seen
Valid until: 1/19/2027
Sources
DeepSeek Open-Sources Inference Framework DeepInfer: The Secret to Halving AI Inference Costs
bilibilicode秘密花园10/21/2026
Related Entities
Related Claims
UnverifiedDeepSeek V4.1 uses local attention for some layers, looking only at the most recent 128 Tokens, and short-term notes are not archived but kept only in local memory and cleared after about three minutes, with the hit rate still above 98%.68% similarUnverifiedDeepSeek V4.1 Flash reduces per-token storage from 3,500 bytes in the previous generation to approximately 890 bytes through KV cache inter-layer alternation and reuse optimization.64% similarUnverifiedTesting shows that DeepSeek V4.1 Flash 1017 reduced its thinking token count by about one-third, producing more concise output.64% similarUnverifiedThe R9700 prefill throughput peaks at 3619 tok/s at a depth of 16000 tokens, drops back to 3192 tok/s at a depth of 64000 tokens, a decay of approximately 12%.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1081777API
curl https://kongchang.com/api/v1/knowledge/claims/1081777MCP
get_claim(id=1081777)