[FEATURE] Support bit and sparsevec types for pgVector

[NikAiyer]

Problem Statement

Currently @mastra/pg only supports vector (full precision) and halfvec (half precision) types. pgvector 0.7.0+ also supports two additional storage types:

1. bit - Binary vectors using PostgreSQL's native bit type, useful for binary quantization which significantly reduces storage and improves search speed
2. sparsevec - Sparse vectors that only store non-zero elements, useful for BM25/TF-IDF representations and other sparse embeddings

Proposed Solution

1. Add bit and sparsevec to the VectorType union type
2. Implement new distance metrics:

• For bit: Hamming distance (<~>) and Jaccard distance (<%>)
• For sparsevec: L2, cosine, and inner product (same as vector)

3. Add appropriate operator classes:
   bit_hamming_ops, bit_jaccard_ops
sparsevec_l2_ops, sparsevec_ip_ops, sparsevec_cosine_ops
4. Implement dimension/element limits:

• bit: up to 64,000 dimensions for indexes
• sparsevec: up to 1,000 non-zero elements for indexes

5. Handle index type restrictions:

• IVFFlat: supports bit with Hamming distance only (bit_hamming_ops), does NOT support Jaccard or sparsevec
• HNSW: supports bit (both Hamming and Jaccard) and sparsevec (L2, cosine, inner product)

Component

RAG

Additional Context

The recent PR #11002 added halfvec support and provides a pattern for implementation

Verification

• I have searched the existing issues to make sure this is not a duplicate
• I have provided sufficient context for the team to understand the request

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and agent prompts for this issue.

• Create Plan

Examples

• Example 1
• Example 2

🔗 Similar Issues

Related Issues

• mastra-ai/mastra#10999

🔗 Related PRs

#10150 - Fix vector definition to fix pinecone [merged]
#10760 - feat(stores): add DuckDB vector store [merged]
#10786 - qualify vector type to fix schema lookup [merged]
#11002 - feat(pg): add halfvec type support for large dimension embeddings [closed]

👤 Suggested Assignees

• abhiaiyer91
• abcfy2
• graysonhicks

---

🧪 Issue enrichment is currently in early access.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

[daneatmastra]

Thank you for reporting this issue! We have assigned it to the trio-tnt and we will look into it as soon as possible.

If you're experiencing an error, please provide a minimal reproducible example whenever possible to help us resolve it quickly.