I Created OctaneDB – The Lightning-Fast Python Vector Database!

#ai #opensource #python #showdev

What is OctaneDB?
OctaneDB is an open-source, high-performance vector database written in Python.
It lets you store, index, and rapidly search millions of text, image, or custom embeddings using state-of-the-art similarity search algorithms.

Key Features
Faster Than Pinecone/ChromaDB: Sub-millisecond queries, >3,000 vectors/sec insert rate.

Advanced Indexing: HNSW for ultra-fast approximate search, FlatIndex for exact matches.

Flexible Storage: In-memory or persistent HDF5 mode.

Text Embedding Built-In: Auto text-to-vector with sentence-transformers.

GPU Acceleration: CUDA support out of the box.

Powerful Search: Batch search, advanced metadata filtering (AND/OR/NOT logic).

Easy Integration: ChromaDB-compatible API for seamless migration.

Open Source: MIT licensed, totally free for all uses!

Try it Online or Locally!
Get Started:

bash
pip install octanedb
Quick Example:

python
from octanedb import OctaneDB
db = OctaneDB(dimension=384, embedding_model="all-MiniLM-L6-v2")
db.create_collection("documents")
db.add(
ids=["doc1", "doc2"],
documents=["About pineapple", "About oranges"]
)
results = db.search_text(query_text="fruit", k=2)
print(results)
Use Cases
Semantic search

NLP & document retrieval

Recommendation engines

Image embedding similarity

RAG pipelines in AI/LLM

Exploratory research

Features Coming Soon
Live Multi-Tenancy

Direct LLM Integration

Hybrid Scalar/Vector Queries

Instant Index Updates (feedback wanted!)

Get Involved!
Try it, star it, and contribute on GitHub

Share your benchmarks and real-world results!

What problems do you face with vector DBs?
Drop your ideas, feature requests, or open an issue!