Model Overview
| Model | Model ID | Dimensions | Context | Max Input |
|---|---|---|---|---|
| Embedding-3 | embedding-3 |
2048/1024/512/256 | 32K | 32K tokens |
| Embedding-2 | embedding-2 |
2048 | 8K | 8K tokens |
Selection Recommendations
- Recommended:
embedding-3— Higher performance, more flexible dimensions
Embedding-3 Dimension Selection
| Dimensions | Recommended Scenario |
|---|---|
| 2048 | High-precision semantic search, RAG |
| 1024 | General retrieval, clustering |
| 512 | Efficient retrieval, classification |
| 256 | Large-scale vector databases, approximate search |
API Key Configuration
From ~/.glm/.env load ZHIPU_API_KEY. Key acquisition:https://open.bigmodel.cn/usercenter/apikeys
API Call
POST https://open.bigmodel.cn/api/paas/v4/embeddings
Supports single and batch text input.
SDK Example
from zai import ZhipuAI
client = ZhipuAI(api_key="your-key")
response = client.embeddings.create(
model="embedding-3",
input=["文本内容"],
dimensions=1024
)
Concurrency Limits
| User Tier | Concurrency |
|---|---|
| V0 | 50 |
| V1 | 100 |
| V2 | 200 |
| V3 | 500 |
Recommended Scenarios
Semantic search, RAG knowledge bases, text clustering, anomaly detection, text classification, similarity computation, recommendation systems.