Vector INSERT Examples - Quick Reference

All Working Syntax Forms

1. Explicit CAST (Your Original Query)

INSERT INTO docs VALUES (1, 'AI', '[0.9, 0.1, 0.0]'::VECTOR(3));

2. Auto-Detection (Recommended - Simpler)

INSERT INTO docs VALUES (1, 'AI', '[0.8, 0.2, 0.0]');

3. With Explicit Column Names

INSERT INTO docs (id, content, embedding)
VALUES (1, 'AI', '[0.9, 0.1, 0.0]'::VECTOR(3));

4. Partial Column List

INSERT INTO docs (id, embedding)
VALUES (1, '[0.9, 0.1, 0.0]');

5. Multiple Rows

INSERT INTO docs VALUES
  (1, 'AI', '[0.9, 0.1, 0.0]'),
  (2, 'ML', '[0.8, 0.2, 0.0]'),
  (3, 'DL', '[0.7, 0.3, 0.0]');

Vector Literal Formats

All these formats work for vector values:

-- With brackets
'[0.9, 0.1, 0.0]'

-- Explicit CAST
'[0.9, 0.1, 0.0]'::VECTOR(3)

-- Different precision
'[0.9234, 0.1567, 0.0123]'

-- Scientific notation
'[9e-1, 1e-1, 0.0]'

Full Example Session

-- Create table
CREATE TABLE docs (
    id INT,
    content TEXT,
    embedding VECTOR(3)
);

-- Insert using CAST
INSERT INTO docs VALUES (1, 'AI', '[0.9, 0.1, 0.0]'::VECTOR(3));

-- Insert using auto-detection (easier!)
INSERT INTO docs VALUES (2, 'ML', '[0.8, 0.2, 0.0]');

-- Query
SELECT id, content FROM docs ORDER BY id;

-- Vector similarity search
SELECT id, content, embedding <-> '[1.0, 0.0, 0.0]' AS distance
FROM docs
ORDER BY distance
LIMIT 5;

Next Steps

Create table with VECTOR column
INSERT vectors (with or without CAST)
Query your data
Create HNSW index for fast similarity search
Run vector similarity queries

Happy testing!