Skip to main content

Current Version

v0.1.0-alpha — Alpha release on Solana devnetLive since: March 2026

What’s Complete

Rust HNSW Core

Production-Ready Query Engine

Status: ✅ Complete
  • 31 unit tests, 37 total tests passing
  • Sub-5ms p99 latency at 100K vectors (384 dims)
  • HNSW graph insert, delete, query, serialize
  • Cosine, Euclidean, Dot Product distance metrics
  • Criterion.rs benchmark suite
  • Zero garbage collection pauses
Location: crates/solvec-core/src/hnsw.rs

Security & Encryption

AES-256-GCM Encryption

Status: ✅ Complete
  • Client-side encryption before storage
  • Keys derived from Solana wallet
  • AES-256-GCM with authenticated encryption
  • VecLabs cannot decrypt user data
Location: crates/solvec-core/src/encryption.rs

On-Chain Verification

Merkle Tree + Proof Generation

Status: ✅ Complete
  • SHA-256 Merkle tree construction
  • 32-byte root hash generation
  • Cryptographic proof generation
  • Proof verification against on-chain root
Location: crates/solvec-core/src/merkle.rs

Solana Integration

Anchor Smart Contract

Status: ✅ Live on devnet
  • Program ID: 8xjQ2XrdhR4JkGAdTEB7i34DBkbrLRkcgchKjN1Vn5nP
  • Example Collection: 8iLpyegDt8Vx2Q56kdvDJYpmnkTD2VDZvHXXead75Fm7
  • 6/6 Anchor tests passing
  • Merkle root storage and retrieval
  • Collection metadata management
View on Explorer: Solana Devnet

SDKs

TypeScript SDK

Status: ✅ Alpha release
  • npm package: solvec@alpha
  • API matches Pinecone for easy migration
  • .upsert(), .query(), .delete(), .verify()
  • Full TypeScript types
npm install solvec@alpha

Python SDK

Status: ✅ Alpha release
  • PyPI package: solvec --pre
  • API matches Pinecone for easy migration
  • .upsert(), .query(), .delete(), .verify()
  • Type hints included
pip install solvec --pre

Work in Progress

Shadow Drive Persistence

Decentralized Vector Storage

Status: 🚧 In progressCurrent: Vectors stored in-memory only (lost on restart)Goal: Encrypted vectors persisted to Shadow Drive with:
  • Automatic sync on write
  • Fast cold-start from Shadow Drive
  • LRU caching for hot vectors
  • Cost: ~$0.000039/MB/epoch
ETA: Late March 2026

WASM Rust Bridge

Browser & Serverless Support

Status: 🚧 In progressCurrent: TypeScript SDK uses JavaScript fallback for HNSW queriesGoal: Compile Rust HNSW core to WebAssembly for:
  • 10x faster queries in browser
  • Serverless Lambda/Cloudflare Workers support
  • Same performance as native Rust
ETA: April 2026

Agent Memory Demo

Reference Implementation

Status: 🚧 In progressGoal: Production-ready demo showing:
  • AI agent with persistent memory
  • Conversation history stored in VecLabs
  • Semantic recall with .query()
  • On-chain verification of agent memory
  • LangChain integration example
Location: demo/agent-memory/ETA: April 2026

Planned Features

Mainnet Deployment

Solana Mainnet Launch

Status: 📋 Planned for Q2 2026Requirements before mainnet:
  • ✅ Anchor program security audit
  • ✅ Shadow Drive persistence complete
  • ✅ 1M+ vector load testing
  • ✅ Beta user feedback incorporated
What changes:
  • Program deployed to mainnet
  • Real SOL required for transactions
  • Production-grade uptime monitoring
  • Enterprise support available

LangChain Integration

Native LangChain Vector Store

Status: 📋 Planned for Q2 2026Goal: Official langchain-solvec package:
from langchain.vectorstores import SolVec
from langchain.embeddings import OpenAIEmbeddings

vectorstore = SolVec(
    wallet="~/.config/solana/id.json",
    collection_name="langchain-memory",
    embedding_function=OpenAIEmbeddings()
)

vectorstore.add_texts(["Agent learned user prefers dark mode"])
results = vectorstore.similarity_search("What are user preferences?", k=5)
Distribution: PyPI + npm + LangChain official docs

AutoGen Integration

Microsoft AutoGen Memory

Status: 📋 Planned for Q2 2026Goal: Native AutoGen memory backend:
from autogen import AssistantAgent
from autogen.memory import SolVecMemory

assistant = AssistantAgent(
    name="assistant",
    memory=SolVecMemory(wallet="~/.config/solana/id.json")
)
Agents automatically store and recall conversation history with on-chain provenance.

Additional Features

Hybrid Search

Combine vector similarity + keyword filters:
collection.query({
  vector: [...],
  filter: { category: 'legal', year: 2026 },
  topK: 10
})

Collection Snapshots

Time-travel queries against historical Merkle roots:
const results = await collection.queryAtTimestamp({
  timestamp: '2026-01-15T00:00:00Z',
  vector: [...]
})

Multi-Tenancy

Isolated collections per user with wallet-based access control

Rust CLI

Command-line tool for devops and testing:
solvec query my-collection --vector-file embedding.json
solvec verify my-collection
solvec export my-collection > backup.jsonl

Version History

v0.1.0-alpha (March 2026)

Initial alpha release on Solana devnet Added:
  • Rust HNSW core with sub-5ms p99 latency
  • AES-256-GCM client-side encryption
  • Merkle tree + proof generation
  • Solana Anchor program (devnet)
  • TypeScript SDK (npm: solvec@alpha)
  • Python SDK (PyPI: solvec --pre)
  • Comprehensive test suite (37 tests)
  • Benchmark suite with Criterion.rs
  • API compatibility with Pinecone
⚠️ Known Limitations:
  • In-memory only (no persistence yet)
  • JavaScript fallback for queries (WASM in progress)
  • Devnet only (mainnet planned Q2 2026)

Contributing to Roadmap

We prioritize features based on user feedback.

Request a Feature

Open a GitHub issue describing your use case

Vote on Features

Upvote existing feature requests in GitHub Discussions

Join Discord

Discuss roadmap priorities with the community

Contribute Code

See CONTRIBUTING.md for development setup

Priority Areas

If you’re interested in contributing, these areas have the highest impact:
Goal: Leverage CPU SIMD instructions for faster distance calculationsImpact: 2-3x query speedup on AVX2/AVX-512 CPUsSkills needed: Rust, SIMD, low-level optimization
Goal: Production-ready persistence layer with smart cachingImpact: Enables production deployments with restart safetySkills needed: Rust, async I/O, Solana Shadow Drive SDK
Goal: Official vector store integrationImpact: Instant distribution to millions of LangChain usersSkills needed: Python, LangChain architecture
Goal: Go, Rust (native), Ruby, Java SDKsImpact: Broader ecosystem adoptionSkills needed: Target language + SDK design

Release Schedule

1

March 2026: Alpha (Current)

✅ Core features complete, devnet live, SDKs published
2

April 2026: Beta

🚧 Shadow Drive persistence, WASM bridge, agent demo complete
3

May 2026: Release Candidate

📋 Security audit, load testing, mainnet deployment prep
4

June 2026: v1.0 Mainnet

📋 Production launch, LangChain/AutoGen integrations, enterprise support

Stay Updated

GitHub Releases

Subscribe to release notifications

Twitter/X

Follow @veclabs for weekly updates

Discord

Join #announcements channel

Questions?

See the FAQ or open a GitHub issue.

Build docs developers (and LLMs) love