Qdrant "Wrong input: Vector dimension error: expected dim, got"

Q: How do I fix "Qdrant "Wrong input: Vector dimension error: expected dim, got""?

1. Compare the two numbers in the error: "expected dim" is the collection size, "got" is your embedding length Open the collection info (GET /collections/ ) and check vectors.size 2. Make the embedding model output match: Set the same model and the same "dimensions" value for indexing and querying Restart your app so cached clients pick up the setting 3. Recreate the collection for the new size: Create a new collection with the new vector size Re-embed and upsert all documents; Qdrant cannot resize an existing collection Delete the old collection once the new one works

Ongoing Self-hosted, Qdrant Cloud Updated November 14, 2026Based on 4 user reportsHow we research

Quick answer

The collection has a fixed vector size and the vector you sent has a different length. Match the embedding model output to the collection size, or recreate the collection and re-embed.

Most likely fix: Compare the two numbers in the error

How to fix it

Try these in order — the most common fix is first.

1. Compare the two numbers in the error
1. "expected dim" is the collection size, "got" is your embedding length
2. Open the collection info (GET /collections/<name>) and check vectors.size
Did this fix it?
2. Make the embedding model output match
1. Set the same model and the same "dimensions" value for indexing and querying
2. Restart your app so cached clients pick up the setting
Did this fix it?
3. Recreate the collection for the new size
1. Create a new collection with the new vector size
2. Re-embed and upsert all documents; Qdrant cannot resize an existing collection
3. Delete the old collection once the new one works
Did this fix it?

Symptoms

Upsert or search returns 400 Bad Request: "Wrong input: Vector dimension error: expected dim: 768, got 1536" (or similar numbers).

Why it happens

Collection created with one size (for example 768) but the embedding model outputs another (1536, 3072)
OpenAI "dimensions" parameter set differently at index time and query time
Embedding model changed after the collection was created
Named vectors: wrong vector name used