models.batchEmbedContents 原生接口。requests 中每个对象都使用与单条向量化相同的字段。为保持 Gemini 原生协议兼容性,model 应填写 models/{model},并与 URL 路径中的模型一致。Authorization: Bearer ********************{
"requests": [
{
"model": "models/gemini-embedding-001",
"content": {
"parts": [
{
"text": "第一段需要向量化的文本。"
}
]
},
"taskType": "RETRIEVAL_DOCUMENT",
"title": "第一段"
},
{
"model": "models/gemini-embedding-001",
"content": {
"parts": [
{
"text": "第二段需要向量化的文本。"
}
]
},
"taskType": "RETRIEVAL_DOCUMENT",
"title": "第二段"
}
]
}curl --location 'https://vip.927ai.ai/v1beta/models/:batchEmbedContents' \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '{
"requests": [
{
"model": "models/gemini-embedding-001",
"content": {
"parts": [
{
"text": "第一段需要向量化的文本。"
}
]
},
"taskType": "RETRIEVAL_DOCUMENT",
"title": "第一段"
},
{
"model": "models/gemini-embedding-001",
"content": {
"parts": [
{
"text": "第二段需要向量化的文本。"
}
]
},
"taskType": "RETRIEVAL_DOCUMENT",
"title": "第二段"
}
]
}'{
"embeddings": [
{
"values": [
0.012345,
-0.023456
]
},
{
"values": [
0.045678,
-0.056789
]
}
]
}