allanpinecone/pinecone-index-arm64 repository overview

⁠Pinecone Local — native arm64 build

An unofficial native linux/arm64 build of Pinecone Local⁠, the in-memory Pinecone emulator for local development and testing.

The official ghcr.io/pinecone-io/pinecone-local image is published for linux/amd64 only, so on Apple Silicon / AWS Graviton it runs under emulation (Rosetta/QEMU). This image is compiled for arm64 and runs natively — no --platform linux/amd64, no emulation.

⁠Images

⁠Quick start

docker run -d \
  --name pinecone-local \
  -e PORT=5080 \
  -e PINECONE_HOST=localhost \
  -p 5080-5090:5080-5090 \
  allanpinecone/pinecone-local-arm64:latest

No --platform flag is needed on ARM. Then point the Pinecone SDK at the local host:

from pinecone import Pinecone
pc = Pinecone(api_key="pclocal", host="http://localhost:5080")

pc.create_index(
    name="example",
    dimension=4,
    metric="cosine",
    spec={"serverless": {"cloud": "aws", "region": "us-east-1"}},
)
idx = pc.Index(host="http://localhost:5081")
idx.upsert(vectors=[{"id": "a", "values": [0.1, 0.2, 0.3, 0.4]}])
idx.query(vector=[0.1, 0.2, 0.3, 0.4], top_k=1)

⁠docker-compose

services:
  pinecone:
    image: allanpinecone/pinecone-local-arm64:latest
    environment:
      PORT: 5080
      PINECONE_HOST: localhost
    ports:
      - "5080-5090:5080-5090"

⁠Limitations

Same as official Pinecone Local:

• In-memory only — data does not persist after the container stops.
• Not for production. Max 100,000 records per index.
• API keys are ignored (no authentication).
• API version 2025-01 (not the latest stable).

⁠Disclaimer

Unofficial, community/internal build provided as-is for local development on ARM. Usage of Pinecone Local is subject to the Pinecone Local EULA⁠.