Unverified50% confidenceBenchmarkExact time
The three models that passed all four tasks were the 3-billion-parameter Granite, Gemma 2B, and Gemma 4B.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/5/2026
First Seen
Valid until: 1/3/2027
Sources
Related Entities
Related Claims
Unverified九个Checkpoint由四个目标模型（Qwen3的4B、8B、14B及Gemma 4的12B）乘以三种算法组成71% similarUnverifiedgemma-4-26b-a4b 模型名称中的 'a4b' 表示该模型采用了混合专家（MoE）架构，激活参数量约为4B67% similarUnverifiedAgent A1 是一个350亿参数的MoE模型，每次推理仅激活约30亿参数65% similarUnverified参数网格C(3值)×gamma(4值)×kernel(3值)×degree(4值)共产生144种参数组合64% similarUnverified4B表示模型参数量为40亿，属于小型语言模型（SLM）范畴63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/977941API
curl https://kongchang.com/api/v1/knowledge/claims/977941MCP
get_claim(id=977941)