Expired85% confidenceFactTime unknown
The Qwen3.6 35B model contains 256 experts, with each token activating only 8 routed experts and 1 shared expert, resulting in approximately 3B activated parameters during actual inference.
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Sources
Related Entities
Related Claims
VerifiedQwen3.6 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3B75% similarUnverifiedQwen Coder adopts a Mixture of Experts (MoE) architecture with 480 billion total parameters, activating only 35 billion parameters during inference.72% similarUnverifiedIn actual testing using the Qwen3 27B Q4 quantized dense model, the generation speed with thinking mode disabled is approximately 58–60 tokens/s.70% similarVerifiedQwen3-235B has a total parameter count of 235 billion, adopts a MoE architecture, and activates approximately 22B parameters during each forward inference pass.70% similarUnverifiedAfter a developer applied abliteration to Qwen3-VL-4B-Instruct, the HarmBench compliance rate reached 100%, compared to only 30.8% for the original model.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/17983API
curl https://kongchang.com/api/v1/knowledge/claims/17983MCP
get_claim(id=17983)