Unverified50% confidenceFactExact time
MiniMax H3 accepts 5120-dimensional text conditional input, while SenseNova's language representations are 4096-dimensional.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMiniMax officially provides prompt specifications for H3, but the document structure is relatively complex and unfriendly to beginners.66% similarUnverifiedThe current Semantic Bridge only applies to the standard MiniMax H3 FL2VA/text-conditioned workflow.65% similarUnverified所有Gemma 4模型原生支持多模态（文本、图像、最多30秒音频输入），上下文窗口为128K Token，支持超过40种语言63% similarUnverified主流嵌入模型如all-MiniLM-L6-v2、BGE系列、E5系列通常基于BERT架构，将文本编码为384至1536维的稠密浮点向量63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/911890API
curl https://kongchang.com/api/v1/knowledge/claims/911890MCP
get_claim(id=911890)