Unverified50% confidenceFactExact time
The V4.1 Flash model contains 384 experts, only a few of which are activated per token; these experts are distributed across 64 cards, 6 per card, plus a clone of one popular chef.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/21/2026
First Seen
Valid until: 1/19/2027
Sources
Related Entities
Related Claims
UnverifiedV4.1 Flash's expert scheduling strategy: load is monitored every 1,000 steps, experts are reallocated every 10,000 steps, and ultimately the busiest card is only 9% busier than the average level.78% similarUnverifiedv4.1 Flash uses a Mixture-of-Experts (MoE) architecture, with each layer containing 384 routed expert networks plus 1 shared expert, and the router selects 6 of them for each token.76% similarUnverifiedQwen3.8 Flash is a MoE model with 24,576 experts across 48 layers internally, activating only approximately 10 experts per generated token.71% similarUnverifiedThe reason V4.1 Flash's total cost increased is that it tends to spawn more sub-Agents in complex tasks; in the benchmark, two problems cumulatively spawned 37 sub-Agents and burned through more than one million tokens.69% similarUnverifiedQwen1.5的MoE版本采用4个共享专家加60个细粒度专家、每次激活4个的配置69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1082314API
curl https://kongchang.com/api/v1/knowledge/claims/1082314MCP
get_claim(id=1082314)