Unverified50% confidenceFactExact time
In MoE architectures, activated parameters determine compute and speed, while total parameters determine memory, so 2B of activation does not make the 14.5 GB Q4_K_M file any smaller.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMoE模型采用稀疏激活，Qwen 3.5 122B总参数122B但每次推理仅激活约10B，需将全部参数加载进内存，对内存容量极度敏感75% similarVerifiedQ4_K_M uses mixed precision, storing some critical weights at 6-bit and the remainder at 4-bit, with an overall average of approximately 4.5-bit.74% similarUnverifiedMoE models read only the activated subset of parameters at each step, making them much faster than dense models of the same parameter count; for example, GLM 5.3 Flash reads only about 30 GB per token.70% similarUnverifiedWithin the Qwen3 series there is a MoE model with a total parameter count of 2.4T and approximately 95B activated parameters, and downloading the model files requires approximately 5TB of storage space.70% similarUnverifiedAlthough MoE models have low computational requirements, the entire model still needs to be loaded into memory, so the total memory requirement is not low; however, memory bandwidth pressure is greatly reduced because only the activated parameters are read each time.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1029964API
curl https://kongchang.com/api/v1/knowledge/claims/1029964MCP
get_claim(id=1029964)