Index_fullness at 100% even with a limited number of vectors

We’ve uploaded approx 341k vectors with 384 dimensions, yet the S1 pod we’re using for testing is showing 100% fullness. Each vector has metadata measured in < 200 bytes.

I’m confused. The pricing page implies that an s1 pod can fit 5M vectors at 786 dimensions, and we are nowhere near 5m vectors and are at half the dimensionality.

Hi @samcfinan. This is because you’re not using an s1 pod; you’re using our free tier, nano pods. Those are designed to hold 100k vectors at 1536 dimensions, so about 340k vectors of 384 dimensions, with metadata, sounds about right.

If you want to make use of s1 please update to our Standard plan and create a new project in one of our non-free regions.

2 Likes

Ah, that makes sense.

Ok, will do. We’re looking to run some realistic benchmarks against an S1 pod and hoped we could do that on the free tier.