v0.x early access

Pricing.
Your cloud, your bill.

polign_db runs in your own cloud, so there is no metered service to pay for. You pay your cloud provider for a bucket and the nodes you choose to run. While a collection sleeps, you pay nothing.

Self-hosted

$0

Free while polign_db is in v0.x early access. No license key, no time limit on what you deploy.

  • The full feature set, with no tiers and no add-ons
  • One static binary; durable with a single -store flag
  • Your bucket: AWS S3, GCS, Azure Blob, MinIO, R2
  • Go package and Python SDK included
  • Community support on GitHub

Production Support

Let's talk!

For teams running polign_db in production who want someone on the hook besides themselves.

  • Priority support with a named contact
  • Deployment and tuning guidance for your workload
  • Upgrade planning and operational review
  • Commercial terms for production use

Every tier runs in your cloud, on your bucket, under your keys. The prices above buy support and partnership, not access: the software itself stays free to self-host during early access. We are committed to reducing what this costs you and improving the experience as polign_db matures.

What running it actually costs

The infrastructure bill is yours, and it is small: these numbers were measured with polign_db's benchmark harness against same-region Amazon S3. Full methodology on the benchmarks page. For a live example, read the case study: 12.5M Wikipedia passages for about $4.60 a month.

~$3.20
per million searches straight from S3, no index in RAM, no idle servers
~$0.01
per million durable writes through the broker-free S3 log
$0
while a collection sleeps. Nodes are disposable and scale to zero
32×
more vectors in the same RAM with the compressed index, 97.4% recall

Point it at your bucket.

Install the server and bring your own S3.