Verified65% confidenceFactExact time
The MoE architecture dynamically selects a small number of experts to participate in computation through a router, achieving sparse activation; mainstream frontier models including GPT-4, Mistral, and DeepSeek all adopt similar designs
3
Sources
65%
Confidence
Long-term
Relevance
7/10/2026
First Seen
Sources
Tencent Hunyuan HY3 Official Version Open-Sourced: The Strongest Under 300B, at 1 Yuan per Million Tokens
bilibiliinfinite灵感港7/6/2026
Related Claims
UnverifiedGPT-4、Mistral等大语言模型已通过MoE架构将不同输入路由至专门化的子网络处理79% similarVerifiedGPT-4和Mixtral等主流大模型也采用了MoE架构78% similarUnverifiedThe MoE architecture can have an extremely large total parameter count while activating only a small fraction of parameters per inference; top-tier models including GPT-4, Mixtral, and Gemini all adopt this architecture.78% similarUnverified分层模型策略将 GPT-4o 等高能力模型用于规划决策节点，中间执行步骤使用 GPT-4o-mini 或本地开源模型（如 Llama 3）以控制成本75% similarUnverifiedThe industry widely believes that GPT-4 incorporates a Mixture of Experts (MoE) mechanism that dynamically routes and activates partial expert sub-networks during each inference74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/465551API
curl https://kongchang.com/api/v1/knowledge/claims/465551MCP
get_claim(id=465551)