> Full Neon documentation index: https://neon.com/docs/llms.txt

### On-disk support for HNSW indexes with pg_embedding

Neon's `pg_embedding` extension, which enables graph-based vector similarity search in Postgres using the Hierarchical Navigable Small World (HNSW) algorithm (HNSW), now persists HNSW indexes to disk. In the previous `pg_embedding` version (0.1.0 and earlier), indexes resided in memory.

Additionally, `pg_embedding` now supports cosine and Manhattan distance metrics.

- Cosine distance

  ```sql
  CREATE INDEX ON documents USING hnsw(embedding ann_cos_ops) WITH (dims=3, m=3, efconstruction=5, efsearch=5);
  SELECT id FROM documents ORDER BY embedding <=> array[3,3,3] LIMIT 1;
  ```

- Manhattan distance

  ```sql
  CREATE INDEX ON documents USING hnsw(embedding ann_manhattan_ops) WITH (dims=3, m=3, efconstruction=5, efsearch=5);
  SELECT id FROM documents ORDER BY embedding <~> array[3,3,3] LIMIT 1;
  ```

Also, be sure to check out the new [Neon AI page](https://neon.com/docs/ai/ai-intro) on our website, and our [docs](https://neon.com/docs/ai/ai-intro), which include links to new [AI example applications](https://neon.com/docs/ai/ai-intro#example-applications) built with Neon Serverless Postgres.