Unverified90% confidenceFactTime unknown
全部四种Gemma 4模型都支持文本和图片输入，Edge模型（E2B和E4B）还额外支持音频输入
1
Sources
90%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified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73% 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66% similarVerifiedThe Ideogram series of models is renowned for its text generation capability, and Ideogram 4 has a noticeably higher success rate in scenarios such as short English sentences, headlines, brand slogans, and poster copy.59% similarVerifiedText encoders such as CLIP and T5 are responsible for converting natural language into vector representations understandable by the model, and are the key modules that determine prompt adherence.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23874API
curl https://kongchang.com/api/v1/knowledge/claims/23874MCP
get_claim(id=23874)