Verified80% confidenceFactTime unknown
Gemma 4's 26B MoE model uses a Mixture of Experts architecture, with 26 billion total parameters but only approximately 4 billion parameters active at any given moment.
6
Sources
80%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe Iris model adopts a Mixture-of-Experts (MoE) architecture, where 35B-A3B indicates a total parameter count of 35 billion and an activated parameter count of 3 billion per inference.78% similarUnverifiedGemma 4 26B-A4B is a MoE model that activates approximately 4B parameters per token, with 26B being the total parameter count and A4B representing approximately 4 billion parameters activated per token.77% similarUnverifiedGemma 4采用混合专家架构（MoE），E2B后缀表示Effective 2B，每次推理只激活约20亿有效参数77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23876API
curl https://kongchang.com/api/v1/knowledge/claims/23876MCP
get_claim(id=23876)