Upstash Vector databases in public beta

We’re delighted to announce public beta availability of Upstash Vector on Fly.io infrastructure! Check our getting started docs to create an index in either the iad and fra regions.

Vector databases are designed for the efficient storage and querying of data based on similarity. They manage vector embeddings—numerical representations of data items in a high-dimensional space—which enable them to perform similarity searches effectively.

These databases have a wide range of uses, from customizing AI responses to semantic image search. A vector database is a superb complement to apps using Fly.io GPUs. Check out Upstash’s collection of example applications and available embedding models to learn more.

While in beta, please avoid running production workloads. Check out the Upstash Vector Roadmap to learn about Upstash’s plans. As General Availability approaches, we’ll post updates on this forum.

Vector is billed per request with no minimum charge. To help you predict costs, you’ll soon see usage details in your Fly.io billing dashboard. We won’t charge you until Vector becomes generally available. You’ll be notified about GA if you have provisioned an index.

We look forward to your feedback!

3 Likes

Will Vector be available in all regions once it becomes GA?

That’s the goal. Is there a specific region you’re interested in right now?

Sydney region.

1 Like

we have just added syd region on your request.

2 Likes

look at that customer service!

1 Like

How strictly should this be followed? Is this more of a “there’s still certification to be done” or more like “you might lose all of your data”?

Vector is a production-grade product - Upstash has already been operating it on AWS. But it’s not running highly-available yet on Fly.io, which could lead to data loss in the worst case scenario.

Upstash is working on HA - we’ll update here when that’s ready!