Skip to main content

​
Migrate from Pinecone

The VecLabs API is intentionally shaped to match Pinecone’s client. For most applications, migration is three line changes. Everything below the client initialization stays identical.

​
The three changes

# Before
from pinecone import Pinecone
pc = Pinecone(api_key="YOUR_KEY")
index = pc.Index("my-index")

# After

from solvec import SolVec
sv = SolVec(network="mainnet-beta") # or "devnet" for testing
index = sv.collection("my-index", dimensions=1536)

# Everything below stays identical ↓

index.upsert(vectors=[...])
index.query(vector=[...], top_k=10)
index.delete(ids=["id1", "id2"])

# New - Pinecone has no equivalent

index.verify() # cryptographic proof on Solana

​
API mapping

PineconeVecLabsNotes
Pinecone({ apiKey })SolVec({ network })No API key needed
pc.Index(name)sv.collection(name, { dimensions })Dimensions required
index.upsert({ vectors })collection.upsert([...])Same structure
index.query({ vector, topK })collection.query({ vector, topK })Same parameters
index.deleteOne(id)collection.delete(id)
index.deleteMany(ids)collection.delete([...ids])
index.fetch(ids)collection.fetch(ids)
index.describeIndexStats()collection.stats()
N/Acollection.verify()Solana Merkle proof

​
Vector format

Pinecone and VecLabs use the same vector record format: 
// Pinecone format
{
  id: 'doc_001',
  values: [0.1, 0.2, ...],
  metadata: { text: 'Hello world' }
}

// VecLabs format - identical
{
  id: 'doc_001',
  values: [0.1, 0.2, ...],
  metadata: { text: 'Hello world' }
}
If you’re already building vector records for Pinecone, you can pass them directly to VecLabs with no transformation.

​
What you gain

After migrating, you get everything Pinecone offers plus:
  • 88% cost reduction - 8/monthvs8/month vs 70/month at 1M vectors
  • Lower latency - 4.7ms p99 vs ~30ms p99 at 1536 dimensions
  • Client-side encryption - VecLabs cannot read your vectors; Pinecone can
  • On-chain audit trail - Merkle root on Solana after every write
  • No API key - your wallet is your credential, not a revocable API key
  • Open source - MIT licensed, fully auditable

​
Full migration example

Here’s a complete before/after for a RAG application: 
from pinecone import Pinecone
from openai import OpenAI

pc = Pinecone(api_key="pc-your-api-key")
index = pc.Index("knowledge-base")
openai_client = OpenAI()

def store_document(doc_id: str, text: str):
response = openai_client.embeddings.create(
model="text-embedding-ada-002",
input=text
)
embedding = response.data[0].embedding

    index.upsert(vectors=[{
        "id": doc_id,
        "values": embedding,
        "metadata": {"text": text}
    }])

def search(query: str, top_k: int = 5):
response = openai_client.embeddings.create(
model="text-embedding-ada-002",
input=query
)
query_embedding = response.data[0].embedding

    results = index.query(
        vector=query_embedding,
        top_k=top_k,
        include_metadata=True
    )
    return results.matches
The diff is exactly three lines. Everything else is unchanged.

​
Migrating existing vectors

If you have existing vectors in Pinecone that you want to bring to VecLabs: 
from pinecone import Pinecone
from solvec import SolVec

# Source
pc = Pinecone(api_key="YOUR_PINECONE_KEY")
pinecone_index = pc.Index("my-index")

# Destination
sv = SolVec(network="devnet")
veclabs_collection = sv.collection("my-index", dimensions=1536)

# Fetch from Pinecone and upsert to VecLabs in batches
batch_size = 100
cursor = None

while True:
    # List vector IDs from Pinecone
    result = pinecone_index.list(limit=batch_size, pagination_token=cursor)
    ids = [v.id for v in result.vectors]

    if not ids:
        break

    # Fetch vectors
    fetched = pinecone_index.fetch(ids=ids)

    # Upsert to VecLabs
    veclabs_collection.upsert([
        {
            "id": id,
            "values": vector.values,
            "metadata": vector.metadata or {}
        }
        for id, vector in fetched.vectors.items()
    ])

    print(f"Migrated {len(ids)} vectors")

    cursor = result.pagination.next if result.pagination else None
    if not cursor:
        break

print("Migration complete.")

# Verify the migration
proof = veclabs_collection.verify()
print(f"Verified on-chain: {proof.verified}")
print(f"Explorer: {proof.solana_explorer_url}")

​
Need help?

Join the Discord and post in #help. Migration questions are always answered within 24 hours.