Verified75% confidenceFactExact time
DeepSeek-V4.1-Flash adopts a MoE (Mixture of Experts) backbone architecture with 552B parameters.
3
Sources
75%
Confidence
Long-term
Relevance
9/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDeepSeek-V4 Flash采用混合注意力、mHC混合机制以及哈希路由的MoE等架构设计78% similarVerifiedDeepSeek V4 Flash is an MoE multimodal model with a total parameter count of 284B.78% similarUnverifiedDeepSeek V4.1 Flash is on par with GLM 5.3 and Kimi K3 on several key benchmark tests.78% similarVerifiedPangu 2.0 Flash adopts a MoE (Mixture of Experts) architecture with a parameter scale of 92B A6B (92B total parameters, 6B activated parameters).77% similarUnverifiedTypical representatives of the MoE architecture include Mixtral 8x7B and DeepSeek-V2.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/921478API
curl https://kongchang.com/api/v1/knowledge/claims/921478MCP
get_claim(id=921478)