Unverified50% confidenceFactExact time
In Unsloth, keeping the Engram table of DeepSeek-V4.1-Flash-1017 on the SSD, the model takes about 125 GB of memory to run under 1-bit quantization and about 300 GB under 4-bit quantization.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
UnverifiedThe 2-bit quantized DeepSeek 763B model takes up about 175GB of memory, can run on a 256GB machine and leaves room for the KV cache.75% similarUnverifiedDeepSeek 1017's expert weights are already FP4 (4 bits per parameter), so conventional 4-bit quantization is of little help; the Engram memory table can be placed on SSD (reading only about 12KB per token), but the expert weights must reside in high-speed memory, because the router selects 6 of 384 experts per layer and makes a different selection for each token.75% 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.74% similarUnverifiedThe hardware used by this user is the Minisforum MS-S1 395+ Max with 128GB unified memory.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1070379API
curl https://kongchang.com/api/v1/knowledge/claims/1070379MCP
get_claim(id=1070379)