Expired75% confidenceFactTime unknown
Gemma 4's E2B version performs comparably to Gemma 3's 27.1-billion-parameter model on multiple tasks in community benchmark tests.
1
Sources
75%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedGemma 4采用混合专家架构（MoE），E2B后缀表示Effective 2B，每次推理只激活约20亿有效参数77% similarVerifiedGemma 4's 26B MoE model uses a Mixture of Experts architecture, with 26 billion total parameters but only approximately 4 billion parameters active at any given moment.73% similarUnverifiedGemma 4's 31B model scored 89.2% on the M2026 evaluation, and the code model's Elo rating is approximately 250072% similarUnverifiedThe Gemma 2B model has approximately 2 billion parameters.72% similarUnverifiedIn evaluations, Gemma 4 31B surpassed models of similar parameter scale such as Zhipu GLM and DeepSeek V3.2, and performed comparably to large models such as Gemini 5 and Kimi 2.5.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23868API
curl https://kongchang.com/api/v1/knowledge/claims/23868MCP
get_claim(id=23868)