Quickstart

Get from zero to your first search result in under 60 seconds.

1. Start Qdrant

Start Qdrant

docker run -p 6333:6333 -p 6334:6334 qdrant/qdrant

Or use a Qdrant Cloud cluster.

2. Connect

Local Qdrant (no inference):

Local

qql-go connect --url http://localhost:6334

Qdrant Cloud:

Cloud

qql-go connect --url https://<cluster>.qdrant.io --secret <api-key>

Local with embeddings (local inference mode):

Local inference

qql-go connect
--url http://localhost:6334
--inference-mode local
--embedding-endpoint http://127.0.0.1:1234/v1/embeddings
--embedding-model text-embedding-all-minilm-l6-v2-embedding

Verify the connection:

Verify

qql-go doctor

3. Create a Collection

Create collection

qql-go exec "CREATE COLLECTION docs HYBRID"

Collection types

The HYBRID keyword creates a collection with both dense and sparse vector support.

With HNSW tuning and quantization:

Advanced collection

qql-go exec "CREATE COLLECTION docs HYBRID WITH HNSW (m = 32, ef_construct = 100) WITH QUANTIZATION (type = 'scalar')"

4. Create Payload Indexes

Always create indexes on fields you'll filter on:

Indexes

qql-go exec "CREATE INDEX ON docs FOR category TYPE keyword" qql-go exec "CREATE INDEX ON docs FOR year TYPE integer"

5. Insert Documents

Insert

qql-go exec "INSERT INTO docs VALUES {'id': 1, 'text': 'Qdrant stores dense and sparse vectors', 'category': 'search'} USING HYBRID" qql-go exec "INSERT INTO docs VALUES {'id': 2, 'text': 'QQL makes vector search feel like SQL', 'category': 'tooling'} USING HYBRID" qql-go exec "INSERT INTO docs VALUES {'id': 3, 'text': 'Hybrid search combines dense and BM25 sparse signals', 'category': 'search'} USING HYBRID"

Batch inserts

You can also batch multiple documents in a single statement for better performance.

Batch insert multiple documents in one statement:

Batch insert

qql-go exec "INSERT INTO docs VALUES {'id': 4, 'text': 'Qdrant supports HNSW indexing'}, {'id': 5, 'text': 'Score boosting with BOOST expressions'} USING HYBRID"

6. Search

Search examples

qql-go exec "QUERY 'vector database' FROM docs LIMIT 5" qql-go exec "QUERY 'vector database' FROM docs LIMIT 5 USING HYBRID" qql-go exec "QUERY 'search' FROM docs LIMIT 5 WHERE category = 'search'"

7. Explain a Query (without executing)

Explain

qql-go explain "QUERY 'vector database' FROM docs LIMIT 5 USING HYBRID RERANK"

8. Explore with the REPL

REPL

qql-go repl

REPL tips

The REPL supports tab completion and multi-line statements.

---

Common Next Steps

Goal	Command
Show all collections	qql-go exec "SHOW COLLECTIONS"
Collection details	qql-go exec "SHOW COLLECTION docs"
Structured JSON output	qql-go exec --quiet --json "SHOW COLLECTIONS"
Browse all points	qql-go exec "SCROLL FROM docs LIMIT 20"
Delete a collection	qql-go exec "DROP COLLECTION docs"

→ Hybrid Search guide
→ Syntax Reference
→ Configuration