Unverified50% confidenceFactExact time
The Kunon 3.8 FlashNext model backbone has 125B parameters, with a total of 24,576 experts; of the 512 routed experts per layer, only 10 are activated per token, so the number of parameters actually involved in computation per pass is about 6B.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Running 125B MoE on a 12GB GPU? Inside Strata's Three-Tier Cooperative Engine
bilibiliNicksxs11/3/2026
Related Entities
Related Claims
UnverifiedThe Kunon 3.8 FlashNext model comes with an Ngram table of about 51B parameters.72% similarUnverifiedQwen3.8 Flash Next has 125 billion total parameters, with only approximately 6 billion parameters activated per token during inference70% similarUnverifiedQwen 3.8 Flash Next is a MoE model with a total parameter count of 125B, but only a small subset of the expert networks is activated per inference.68% similarUnverifiedThe author tested Qwen3 Flash Next using dual 7900 XTX graphics cards, Qwen3 Q3_K_XL quantization, and a 32K context, achieving capability scores of 88 in knowledge reasoning, 100% in instruction following, 30% in advanced mathematics, 90 in Chinese, and 80 in code.67% similarUnverifiedLFM2.5-24B-A2B has 24 billion total parameters and uses an MoE architecture with 64 experts per block, Top-4 dynamic routing, and only 2.45 billion parameters activated per token.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1134844API
curl https://kongchang.com/api/v1/knowledge/claims/1134844MCP
get_claim(id=1134844)