Unverified50% confidenceFactExact time
HY3 internally contains 192 experts and uses a Top-8 routing mechanism.
1
Sources
50%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
腾讯HY3开源模型深度实测：轻量MoE硬刚旗舰级性能
bilibiliAI_Express7/8/2026
Related Claims
UnverifiedHY3's Top-8 routing activation ratio is approximately 7% (21 billion / 295 billion).62% similarUnverifiedKimi K2.7 Code uses a MoE architecture with 384 experts, selecting 8 experts plus 1 shared expert per token, with actual computation approximately 2.3% of that of a dense model.58% similarUnverifiedHy3 adopts a Mixture of Experts (MoE) architecture with a total parameter count of 295 billion (295B), activating only 21 billion (21B) parameters per inference.54% similarUnverifiedRovo Agent ranked first on the SWE-bench full benchmark leaderboard, achieving a resolution rate of 41.98% across 2,300 tasks.54% similarUnverifiedQwen3.6 35B模型包含256个专家，每个Token仅激活8个路由专家和1个共享专家，实际推理时激活参数约为3B53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/606568API
curl https://kongchang.com/api/v1/knowledge/claims/606568MCP
get_claim(id=606568)