Verified65% confidenceFactExact time
MoE models activate only a small fraction of total parameters per inference; Qwen3 Max may activate between 1/8 and 1/16 of the total parameters per inference.
3
Sources
65%
Confidence
Long-term
Relevance
7/23/2026
First Seen
Sources
Related Claims
UnverifiedThe MoE architecture activates only a small fraction of the 2.4T parameters during inference (typically 10–20%).79% similarUnverifiedMoE模型中'35B'指总参数量，每个token只激活少量Expert（通常2-4个），实际计算量可能仅相当于6-7B的Dense模型78% similarUnverified千问3.6的35B MOE模型每次只激活其中3B的参数78% similarPartially VerifiedBoth Qwen 3 Max and DeepSeek-V3 adopt the MoE architecture with total parameter counts reaching hundreds of billions, but the parameters actually activated during inference are only a small fraction of the total.77% similarUnverifiedKolibri-1 has a total parameter count of 78B, with only approximately 3.46B parameters activated during inference.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/602608API
curl https://kongchang.com/api/v1/knowledge/claims/602608MCP
get_claim(id=602608)