VectorField

from django.db import models
from milvusql_django.fields import VectorField

class Item(models.Model):
    category = models.CharField(max_length=64)
    embedding = VectorField(dim=768)

A standard Django Field subclass — not a fake Column shim. It only needs to round-trip list[float] <-> VECTOR(n) DDL text; milvusql's DBAPI already carries the Python list through as a real bind value, so there's no encoding to do on the Python side either (get_prep_value/ to_python/from_db_value are all effectively pass-through).

item = Item.objects.create(category="book", embedding=[0.1] * 768)
item.embedding
# [0.1, 0.1, ..., 0.1]

dim

VectorField(dim=768)

Renders as VECTOR(768) in DDL. Omitting dim renders a bare VECTOR — allowed by MilvusQL's grammar, but Milvus itself requires a fixed dimension for a FLOAT_VECTOR field in practice, so in almost every real model you'll want to set it.

Querying

VectorField doesn't add .filter()-level distance lookups — Django's ORM has no expression for "nearest neighbor," so there's nothing to add a lookup for. Vector search goes through vector_search() instead.