Unverified50% confidenceFactExact time
All three tested models are MOE architectures at the 30B parameter level: Gemma 4 has 26B total parameters with 4B active, GLM 4.7 Flash has 30B total parameters with 3B active, and Qwen 3.6 35B A3B has 35B total parameters with 3B active.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/4/2026
First Seen
Valid until: 2/2/2027
Sources
Bilibili AI Infinite Arena: Local Deployment Extreme PK of Three MOE Models
bilibili九物AI生产力11/4/2026
Related Entities
Related Claims
UnverifiedQwen 3.6's 35B MoE model activates only 3B of its parameters at a time.79% similarUnverified混元3是混合专家模型（MoE），总参数量为295B，激活参数约21B78% similarVerifiedQwen3.6 35B-A3B是混合专家架构（MoE），总参数量35B，每次推理实际激活参数量约3B76% similarUnverifiedPACE was evaluated on four hybrid models, with parameter sizes ranging from 4B to 35B.72% similarUnverifiedTencent open-sourced the Hy-Memory-30B-A3B model, a 30B-parameter mixture-of-experts (MoE) model that activates about 3B parameters per token.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1134817API
curl https://kongchang.com/api/v1/knowledge/claims/1134817MCP
get_claim(id=1134817)