Vector Bucket Local Development
Develop and test vector bucket integrations in your local environment with the Supabase CLI.
This feature is in alpha
Expect rapid changes, limited features, and possible breaking updates. Share feedback as we refine the experience and expand access.
You can now develop and test Vector Bucket integrations in your local environment using the Supabase CLI.
This allows you to build and iterate on your vector search applications without needing to deploy to a live environment. Make sure you have the latest version of the Supabase CLI installed to access this feature.
Local driver
Local vector buckets use pgvector as their storage engine. Hosted vector buckets use Amazon S3 Vectors, so query behavior and performance can differ between environments.
Hosted queryVectors requests accept topK values up to 10,000 and return results in pages of at most 100. Use nextToken to retrieve each additional page. Local requests accept topK values up to 100 and reject nextToken.
See Vector Bucket limits for the full comparison.
Setting up local vector buckets#
Make sure you have the feature enabled in your config.toml file:
# Store vector embeddings in S3 for large and durable datasets[storage.vector]enabled = trueDeclarative configuration#
You can define your vector buckets in the config.toml file using the following syntax:
[storage.vector.buckets.documents-openai][storage.vector.buckets.images]Then use supabase seed buckets to create the buckets in your local environment or linked project.